2004-10-20 Michael Daum * grammar/negra/Makefile (1.9), grammar/penn/Makefile (1.8), utils/Makefile (1.22), utils/depsy.pl (1.10), utils/tagger.pl (1.5), Makefile (1.84): last minute fixes: - install depsy plugins even if the grammars (negra, wsj) are not compiled and distributed - note in a manpage that this is part of the CDG suite; otherwise one might get confused about w_t_f this manual is talkin' about 2004-10-20 Michael Daum * grammar/deutsch/Makefile (1.11): - don't generate a link if it is already there: fixes make dist-check 2004-10-20 Michael Daum * bindings/CDG/Makefile (1.7), bindings/Makefile (1.11), doc/Makefile (1.16), grammar/Makefile (1.5), grammar/stellingen/Makefile (1.34), include/Makefile (1.6), include/config.mk (1.18), libcdg/Makefile (1.60), libcontainer/Makefile (1.35), utils/Makefile (1.21), Makefile (1.83), configure (1.16), yada/Makefile (1.6): - added DESTDIR support: no DESTDIR no packaging 2004-10-20 Michael Daum * doc/make.help (1.2): - added DESTDIR support - clarified usage of dist-tar 2004-10-20 Michael Daum * debian/changelog (1.4): - initial release is _today_ 2004-10-20 Michael Daum * debian/control (1.4): - added build dependecies to tetex-bin and doxygen 2004-10-20 Michael Daum * debian/rules (1.4): - added DESTDIR support - added make variable CONFIGURE_ARGS defaulting to "--prefix /usr --disable-wsj --disable-negra" 2004-10-20 Kilian A. Foth * utils/ccc-renumber (1.19): Handle malformed .cda files (i.e., do not choke and denounce them instead). 2004-10-20 Kilian A. Foth * utils/ccc-annotate (1.4): - Option -a loads all available auto- trees (do NOT use this in ccc/heiseticker...) - understand directory boundaries when using .lastnumber 2004-10-20 Kilian A. Foth * utils/ccc-show-conflicts (1.9): Do not assume that xcdgclient.tcl is in the PATH. 2004-10-20 Kilian A. Foth * utils/tokenize.pl (1.42): Detect words with embedded closing parens. 2004-10-20 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.92), grammar/deutsch/Grammatik.cdg (1.169), grammar/deutsch/Hierarchien.cdg (1.49), grammar/deutsch/Lexikon.cdg (1.130), grammar/deutsch/Namen.txt (1.73), grammar/deutsch/Nomen.txt (1.118), grammar/deutsch/Verben.txt (1.130), grammar/deutsch/doc.tex (1.58), grammar/deutsch/make-verbs.pl (1.49): - random new lexicon items - add more restrictions on several types of nonprojectivity - detect more missing commas following embedded constituents - assume that `immer' + comparative isa constituent - punish simultaneous subclauses at verb and noun - punish adjacent parallel subclauses - demand overt marking of datives (like for genitives) 2004-10-19 Michael Daum * grammar/stellingen/Makefile (1.33): - oops, removed personal extras from dist-files 2004-10-19 Michael Daum * VERSION (1.2): - get me your last-minute "bug fixes" NOW before releasing 0.95 tonight. 2004-10-19 Michael Daum * Makefile (1.82): - oops, dist-files and dist-src depend on a proper ChangeLog 2004-10-19 Michael Daum * grammar/deutsch/doc.tex (1.57): - removed usepackage times and fontencoding 2004-10-19 Michael Daum * bindings/CDG/Makefile (1.6), bindings/Makefile (1.10), cdgp/Makefile (1.32), doc/Makefile (1.15), grammar/berlin/Makefile (1.5), grammar/deutsch/Makefile (1.10), grammar/negra/Makefile (1.8), grammar/penn/Makefile (1.7), grammar/stellingen/Makefile (1.32), include/Makefile (1.5), include/config.mk (1.17), libcdg/Makefile (1.59), libcontainer/Makefile (1.34), test/Makefile (1.21), utils/Makefile (1.20), xcdg/Makefile (1.7), Makefile (1.81), configure (1.15), yada/Makefile (1.5), yada/images/Makefile (1.4): - added support for building source distributions: make dist-src, every Makefile has a variable DIST_FILES that specifies the content of the source tarball; type make dist-files to get a file called "dist-files" that lists them - packaging the ccc/Beispiele into the distribution now - added support for checking a source tarball: make dist-check - added make help: make help - added make variables debugger: make status - added --disable-[wsj|negra|ccc|treetagger] to the configure script to skip the refering checks - fixed/simplified installation of cdgsetup.pl 2004-10-19 Kilian A. Foth * libcdg/input.c (1.307): Allow uppercase after ? and ! (which might actually occur within sentences if the tokenizer so decides). 2004-10-19 Michael Daum * doc/make.help (1.1): - help file for "make help" and "make morehelp" 2004-10-19 Michael Daum * grammar/deutsch/Phänomene.cdg (1.6): - the Phänomene has been superseeded by the Beispiele in ccc 2004-10-18 Michael Daum * doc/db.tex (1.6), doc/doc.tex (1.88), doc/genetic.tex (1.2), doc/libcdg.tex (1.93): - removed T1 fontencoding and usepackage times to get nice pdf fonts 2004-10-18 Michael Daum * include/Makefile (1.4), include/config.mk (1.16): - fixed install/uninstall of the cdgsetup.pl thingy kludge 2004-10-18 Michael Daum * include/config.mk (1.15): - make "make uninstall" a litte quieter 2004-10-18 Michael Daum * grammar/Makefile (1.4): - on "make install" a sym link Beispiele -> deutsch is generated in the GRAMMAR_DIR to satisfy the cdg wrapper script so after the installation you can have cdg deutsch the Deutsch grammar without any examples and cdg Beispiele the Deutsch grammar including all annotated examples 2004-10-18 Michael Daum * grammar/deutsch/Makefile (1.9): - on "make install" Beispiele.cdg is installed using the Deutsch grammar via a load pragma, i.e. Beispiele is not a stand-alone-grammar, you need the Deutsch grammar to play with it. 2004-10-18 Michael Daum * Makefile (1.80): - ChangeLog can only be generated if you are logged in on a nats machine with access to the CVS repo 2004-10-18 Michael Daum * utils/cdg (1.31): - fixed bindings detection: shared libraries aren't executable - added error messages when cdgp, tclcdg.so and xcdg can not be found 2004-10-18 Michael Daum * utils/Makefile (1.19): - added xcdg to install/uninstall - added cdg to uninstall - added xcdg to the all and clean target: generates a link to cdg 2004-10-18 Michael Daum * libcdg/Makefile (1.58): - added parser.output to clean target 2004-10-18 Michael Daum * grammar/deutsch/Makefile (1.8), grammar/negra/Makefile (1.7), grammar/penn/Makefile (1.6): - removed dependency on cdgsetup.pl: this was a bad idea cus every configure will force a grammar rebuild -> too much automatism, I think 2004-10-18 Michael Daum * xcdg/Makefile (1.6): - deprecating installation of xcdg (generated via sed from xcdg.sh) 2004-10-18 Michael Daum * xcdg/xcdgrc.sample (1.4): - this way of starting xcdg is deprecated 2004-10-18 Kilian A. Foth * README (1.9): Recommend `xcdg' instead of `cdg -x'. 2004-10-18 Kilian A. Foth * README (1.8): Fix spelling and numbering. 2004-10-18 Kilian A. Foth * xcdg/parsetree.tcl (1.135): parsetree::highlight{}: Detect malformed highlight specifications and complain instead of throwing an exception. 2004-10-15 Michael Daum * yada/YadaExperiment.tcl (1.11), yada/YadaExperiments.tcl (1.9), yada/YadaGrammar.tcl (1.24), yada/YadaGrammars.tcl (1.26), yada/YadaMachine.tcl (1.13), yada/YadaMachines.tcl (1.12), yada/YadaProc.tcl (1.22), yada/experiments.tcl.sample (1.1), yada/grammars.tcl.sample (1.6), yada/machines.tcl.sample (1.4), yada/methods.tcl.sample (1.7): - uuh, get things running again: don't mind all these changes 2004-10-15 Michael Daum * yada/Process.tcl (1.16): - processes don't run w/o output: why 2004-10-15 Michael Daum * utils/xml2anno.tcl (1.3): - removed hard dependency on tdom 0.5: where are at 0.7.8 now 2004-10-15 Michael Daum * include/Makefile (1.3), include/config.mk (1.14): - link cdgsetup.pl from /lib/cdg/cdgsetup.pl to whereever it is needed 2004-10-15 Michael Daum * grammar/negra/Makefile (1.6): - annotations are up to s19620 - negra.cdg depends on Sätze.cdg 2004-10-14 Michael Daum * utils/cdg (1.30): - further CDGHOME fixes: CDGSYSTEM is the install prefix of cdg _if it is installed_ never mix it with CDGHOME which is _ the users version in his HOME directory_ 2004-10-14 Michael Daum * xcdg/Makefile (1.5): - installing xcdgclient now 2004-10-14 Michael Daum * utils/cdg (1.29): - fixed reference to installed tclbindings 2004-10-14 Michael Daum * utils/ccc-show-conflicts (1.8): - removed absolute path to xcdgclient.tcl: xcdgclient must be in your path now 2004-10-14 Michael Daum * utils/Makefile (1.18): - fixed installation of the cdg wrapper script 2004-10-14 Michael Daum * include/config.mk (1.13): - added tcl sed stripper 2004-10-13 Michael Daum * utils/indexer.c (1.11): - made error reporting print on stderr consistently - quietened the indexer which speeds up things 2004-10-13 Michael Daum * utils/ccc-Makefile (1.12): - fixed usage of DE variable 2004-10-13 Michael Daum * include/config.mk (1.12): - fixed %.dvi rule: latex has to be run twice, no matter what. 2004-10-13 Michael Daum * grammar/deutsch/Makefile (1.7): - fixed installation: now the grammar source files (*.txt etc) are installed also as the ccc suite needs them - added new target clean-docu to clean the docu only 2004-10-13 Michael Daum * xcdg/allwordgraphs.tcl (1.70): Fixed removing the function declaration from Allwordgraphs class. Now the inherrited function (Databrowser::refreshid) is called as it should. 2004-10-12 Michael Daum * doc/doc.tex (1.87): - removed notion about CDGHOME 2004-10-12 Michael Daum * Makefile (1.79): - added clean-debian target to get rid of the debian compile after a make dist-deb 2004-10-12 Michael Daum * README (1.7): - overhaul due to new install system 2004-10-12 Kilian A. Foth * utils/make-extra-lexicon.pl (1.5): Find grammar/deutsch by reasoning from $0 instead of $ENV{CDGHOME}. 2004-10-12 Kilian A. Foth * utils/ccc-check (1.18): Use plain `cdg' instead of $CDGHOME/cdgp/cdgp. 2004-10-11 Michael Daum * debian/README.debian (1.2), debian/changelog (1.3), debian/control (1.3), debian/copyright (1.2), debian/dirs (1.3), debian/rules (1.3), Makefile (1.78): - reworked the debian packageing _completely_ - added a dist-rpm target using alien (I know this does not work on your SUSE distribution) so now we have * make dist-tar to get a tarball containing the binaries that install in PREFIX. Use that variable to specify the target (e.g. make dist-tar PREFIX=/opt/pkg/cdg-0.95 * make dist-deb to get a propper debian package * make dist-rpm to get an rpm packakge which is a kludge as long as there's no real spec file and the package gets build on a SUSE distro * make install to install things directly, complemented by make uninstall. TODO: make install should generate an uninstall- script because you might want to throw away your build environment ... which includes your make uninstall. 2004-10-11 Michael Daum * grammar/negra/Makefile (1.5), grammar/negra/find-gold.sh (1.4): - fixed finding the gold: this got broken by moving the negra gold into ccc 2004-10-11 Michael Daum * Makefile (1.77): - improved cleaning 2004-10-11 Michael Daum * include/config.mk (1.11): - added remote targets for libCDG.so and tclcdg.so 2004-10-11 Michael Daum * configure (1.14): - added check for the CDG Corpus Collection 2004-10-11 Michael Daum * grammar/penn/Makefile (1.5): - remove the target if the generator fails 2004-10-11 Michael Daum * grammar/deutsch/Makefile (1.6): - the generated cdgs depend on the cdgsetup.pl - if the generators fail erase the target 2004-10-11 Michael Daum * bindings/Makefile (1.9): - libCDG.so and tclcdg.so depend on the core libs 2004-10-11 Michael Daum * debian/files (1.2): - this file is generated automatically 2004-10-11 Michael Daum * libcdg/Makefile.am (1.32): - removed leftover from abandoned autotools make system 2004-10-11 Michael Daum * bindings/CDG/Makefile (1.5): - CDG::Utils.1 depends on Utils.1 ;) 2004-10-11 Michael Daum * bindings/CDG/Makefile (1.4): - installing CDG::Utils.1 instead of Utils.1 into the manual namespace 2004-10-11 Michael Daum * utils/Makefile (1.17): - install forgotten cdgsetup.pl s 2004-10-11 Michael Daum * Makefile (1.76): - added make dist-tar call it with make dist-tar PREFIX=/opt or make dist-tar PREFIX=/usr/local to get a binary tarball that runs there 2004-10-11 Michael Daum * include/config.mk (1.10): - fixed installation of cdgsetup.pl 2004-10-11 Michael Daum * utils/Makefile (1.16): - fixed installation of utils: forgotten relinking, forgotten permissions 2004-10-11 Michael Daum * utils/Makefile (1.15): - forgotten RPATH variable during linking of compiletime utils 2004-10-11 Michael Daum * include/config.mk (1.9): - realy fixed make distclean 2004-10-11 Michael Daum * configure (1.13): - generating a reconfigure script that captures the configure args 2004-10-11 Michael Daum * Makefile (1.75): - fixed make distclean 2004-10-11 Michael Daum * bindings/CDG/Makefile (1.3), bindings/Makefile (1.8), cdgp/Makefile (1.31), doc/Makefile (1.14), grammar/deutsch/Makefile (1.5), grammar/negra/Makefile (1.4), grammar/Makefile (1.3), grammar/penn/Makefile (1.4), grammar/stellingen/Makefile (1.31), include/config.mk (1.8), libcdg/Makefile (1.57), libcontainer/Makefile (1.33), test/Makefile (1.20), utils/scripts/Makefile (1.1), utils/Makefile (1.14), utils/cdg (1.28), utils/make-extra-lexicon.pl (1.4), xcdg/Makefile (1.4), xcdg/busy.tcl (1.24), xcdg/startup.tcl (1.7), xcdg/visparses.tcl (1.52), xcdg/xcdg.sh (1.9), xcdg/xcdg.tcl (1.92), Makefile (1.74), yada/Makefile (1.4), yada/yada.sh (1.15), yada/images/Makefile (1.3): - added "install" and "uninstall" targets to the make system 2004-10-11 Michael Daum * grammar/penn/pennplugin.pl (1.10): - some podding for penn aswell 2004-10-11 Michael Daum * grammar/negra/negraplugin.pl (1.75): - fixed pod docu: good podding makes good manuals 2004-10-11 Michael Daum * xcdg/allconstraints.tcl (1.33), xcdg/allfiles.tcl (1.65), xcdg/allhierarchies.tcl (1.34), xcdg/alllevels.tcl (1.37), xcdg/alllexemes.tcl (1.31), xcdg/allnetworks.tcl (1.59), xcdg/allparses.tcl (1.63), xcdg/allwordgraphs.tcl (1.69), xcdg/databrowser.tcl (1.13): - fixed "pre-last line missing" bug - fixed "hierarchies initially empty canvas" bug (the databrowser family needs a refactoring to make any sense at all) 2004-10-08 Kilian A. Foth * utils/cdg (1.27): Prefer the static cdgp over the dynamic if both are present. The static binary Just Works (TM), even across GLIBC versions, the dynamic doesn't. If you don't want the static binary, don't make it. 2004-10-08 Kilian A. Foth * libcdg/write.c (1.66), bindings/CDG/Utils.pm (1.10): Make both C and Perl write annotations with no unnecessary ' around strings. You may laugh, but it does save lots of disk space with huge tree collections. 2004-10-05 Kilian A. Foth * utils/deutsch-tagger.pl (1.17), grammar/deutsch/Adjektive.txt (1.91), grammar/deutsch/Grammatik.cdg (1.168), grammar/deutsch/Hierarchien.cdg (1.48), grammar/deutsch/Lexikon.cdg (1.129), grammar/deutsch/Namen.txt (1.72), grammar/deutsch/Nomen.txt (1.117), grammar/deutsch/TODO (1.10), grammar/deutsch/Verben.txt (1.129), grammar/deutsch/deutsch.m4 (1.7), grammar/deutsch/doc.tex (1.56), grammar/deutsch/extra.m4 (1.15): - random new lexicon items - use entire sentences for examples where possible - demand agreement between NP connected by `als' - forbid `Ich sehe Jane zu laufen' - forbid extraposition of subclauses from vorfeld to mittelfeld - allow `von...bis' to co-occur in vorfeld position - close major loophole: intransitive verbs cannot form normal passives - disprefer adverbials under ATTR, unless marked by preceding DET - detect BILD-style conjunction elision: `Er fluchte, ging weg.' - treat TRUNC subjects as person=third 2004-10-05 Kilian A. Foth * utils/ccc-annotate (1.3): Remember the last number that we started XCDG with, so the next invocation won't have to do quite as much disk I/O to find the next missing tree. 2004-10-05 Michael Daum * configure (1.12): - added check_install 2004-10-05 Kilian A. Foth * xcdg/allparses.tcl (1.62): Fix size and content of several table columns. 2004-10-05 Kilian A. Foth * xcdg/parsetree.tcl (1.134): Make parsetree canvas scroll vertically on mouse wheel events. 2004-10-05 Kilian A. Foth * configure (1.11): Fix db40 check so that it actually notices that it succeeded. 2004-10-04 Michael Daum * bindings/Makefile (1.7), cdgp/Makefile (1.30), include/config.mk (1.7), libcdg/Makefile (1.56), libcdg/command.c (1.309), libcdg/command.h (1.56), libcdg/compile.c (1.87), libcontainer/Makefile (1.32), test/Makefile (1.19), utils/Makefile (1.13), configure (1.10): - added better default options for the constraint compiler cflags etc. - added configure options to set your own custom cflags - renamed INCLUDE to INCLUDES - cdgp and bindings know how to build libblah.so and libcdg.so now: you can simply enter cdg/cdgp and type make to get the cdgp only - minor cleanups in check_bdb - abandoned the idea of generating a versioninfo.c file: this info is provied by autoconfig.h reporting on the version plus the build now. 2004-10-04 Kilian A. Foth * xcdg/parsetree.tcl (1.133): Mouseless editing: correctly handle release events occurring before the press event was fully handled. Problem: the sequence causes trouble when executed quickly. causes xcdg to compute the best label at the current position, and causes the computed value to be applied to the parse. But if the release happens before the optimization has terminated, the current label is applied, and when the optimization terminates its result is discarded. Worse yet, the canvas widget will reflect the new label but the parse will still carry the old one! This is because the release handler cannot wait for the optimization to terminate, it *must* do its work before the other handler can proceed. (Blame Ousterhout's bias for event handling over multithreading for that.) Solution: the Release handler detects that it was called prematurely, sets a flag, and exits; and the Enter handler reads this flag and then knows to call the Release handler again. 2004-10-04 Kilian A. Foth * utils/ccc-check (1.17): Do not suppress error messages if `make' failed (which it can do for a distressing numberof reasons). 2004-10-04 Kilian A. Foth * utils/ccc-distribute (1.2): This script can run a LONG time. Use mv's -v flag to give progress indication. 2004-10-04 Kilian A. Foth * utils/ccc-renumber (1.18): Try harder not to reorder sentences gratuitously, even when two successive sentences have the same reading and you deleted one earlier. This is important for the corpus `Beispiele', where two trees about the same sentence regularly follow each other, one correct and one oncorrect, and the correct one should always come first. That property should not be destroyed by automatic renumbering. 2004-10-04 Kilian A. Foth * utils/ccc-search (1.12): Do not die upon reading undefined feature values, pretend their value is `' like in the old version. 2004-10-04 Kilian A. Foth * utils/ccc-unparse (1.1): Kill all auto-*cda that you have already revised in one stroke! Handle with care! 2004-10-04 Kilian A. Foth * utils/corpus-error.el (1.2): More movement commands, automatic visiting of an offending word, pre-defined elisp functions such as `diff-negra-errors'. 2004-10-04 Kilian A. Foth * utils/cdg (1.26): cdg/cdgp/cdgp is no longer a magic libtool script, so we can use gdb directly again when -d is requested. 2004-10-04 Michael Daum * doc/Makefile (1.13), doc/doc.tex (1.86), grammar/deutsch/Makefile (1.4): - docu is in pdf by default now 2004-10-04 Michael Daum * configure (1.9): - added check_tnt for Torsten Brants' Trigram'n'Tags Tagger - added check_treetagger for Helmut Schmid's tree-tagger 2004-10-04 Kilian A. Foth * xcdg/balloon.tcl (1.9): Do not post a background job for switching on a tooltip when the user requested immediate popup, just do it. 2004-09-30 Michael Daum * bindings/Makefile (1.6), cdgp/Makefile (1.29), libcdg/Makefile (1.55), libcontainer/Makefile (1.31), test/Makefile (1.18), Makefile (1.73), utils/Makefile (1.12): - cleaning does not depend on built dependencies any more - only the top level distclean calls the clean target now: this prevents the clean target to be called multiply 2004-09-30 Michael Daum * include/config.mk (1.6), configure (1.8): - cleaned up the berkley db check to match the other tests 2004-09-30 Kilian A. Foth * configure (1.7), bindings/Makefile (1.5), cdgp/Makefile (1.28), include/config.mk (1.5): Makefile system fixes: - the BDB link option can be different from "-ldb", must be set by configure. - since config.mk is evaluated after autoconfig.mk, it must use += or ?= instead of = for things that autoconfig.mk might change. 2004-09-29 Kilian A. Foth * xcdg/parsetree.tcl (1.132), xcdg/visparses.tcl (1.51): Bind and to "forward/backward canvas" in tree editor. 2004-09-29 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.48): Do NOT assume that all German verbs are all-lowercase; see /data/corpus/ccc/heiseticker/000/heiseticker-s13515.cda. 2004-09-29 Kilian A. Foth * grammar/deutsch/make-names.pl (1.19): Allow declaration of `New York' via: Region New_York 2004-09-29 Kilian A. Foth * grammar/deutsch/Makefile (1.3), grammar/penn/Makefile (1.3), include/config.mk (1.4): Generate cdgsetup.pl everywhere it's needed. Return 1 from cdgsetup.pl so it won't spook the `require'. 2004-09-28 Michael Daum * configure (1.6): - added workaround for buggy tcl installations 2004-09-28 Michael Daum * bindings/Makefile (1.4): - fixed usage of TCL_XXX variables 2004-09-28 Michael Daum * include/config.mk (1.3): - be sensitive to settings guested by configure 2004-09-28 Michael Daum * yada/images/Makefile (1.2), grammar/stellingen/Makefile (1.30): - improved robustness 2004-09-28 Michael Daum * configure (1.5): - added check_tcl 2004-09-28 Michael Daum * Makefile (1.72): - added directory precedence for utils and grammar to include to get cdgsetup.pl in time 2004-09-28 Michael Daum * configure (1.4): - added check_swig 2004-09-28 Michael Daum * bindings/CDG/Makefile (1.2), bindings/Makefile (1.3), cdgp/Makefile (1.27), doc/Makefile (1.12), grammar/berlin/Makefile (1.4), grammar/deutsch/Grammatik.cdg (1.167), grammar/deutsch/Makefile (1.2), grammar/deutsch/make-adjectives.pl (1.33), grammar/deutsch/make-names.pl (1.18), grammar/deutsch/make-nouns.pl (1.26), grammar/deutsch/make-verbs.pl (1.47), grammar/negra/Makefile (1.3), grammar/negra/auto2silver.pl (1.4), grammar/negra/compare.sh (1.17), grammar/negra/find-unknown-names.pl (1.3), grammar/Makefile (1.2), grammar/penn/Makefile (1.2), grammar/penn/make-grammar.pl (1.3), include/Makefile (1.2), include/config.h (1.2), include/config.mk (1.2), libcdg/Makefile (1.54), libcdg/input.c (1.306), libcontainer/Makefile (1.30), test/Makefile (1.17), utils/Makefile (1.11), utils/ccc-renumber (1.17), utils/ccc-search (1.11), utils/cdgc.sh (1.4), utils/extract-grammar.pl (1.3), utils/make-extra-lexicon.pl (1.3), utils/merge-annos.pl (1.6), utils/shift-reduce.pl (1.15), utils/splice-annotations.pl (1.4), xcdg/Makefile (1.3), Makefile (1.71), configure (1.3), yada/Makefile (1.3): - abandoned sed-ing perl files and introduced the cdgsetup.pl trick: all perl scripts look for a file cdgsetup.pl in the same directory the are in and source it; cdgsetup.pl is generated by the make system in cdg/include by default and wherever it is needed (like in cdg/utils); on install time (not implemented yet) a new cdgsetu.pl will be generated that points to the proper install location - configure checks for Term::ReadLine now and chokes if it is not found 2004-09-27 Michael Daum * bindings/Makefile (1.2): - renamed CAT to DEPS 2004-09-27 Michael Daum * yada/Makefile (1.2), xcdg/Makefile (1.2): - fixed make tclIndex 2004-09-27 Michael Daum * configure (1.2): removed version conditional 2004-09-27 Michael Daum * VERSION (1.1): - this file defines the current cdg version 2004-09-27 Michael Daum * grammar/negra/Makefile (1.2): - fixed make check 2004-09-27 Michael Daum * grammar/negra/negra.m4 (1.25): oops including base.cdg using m4 is a bad idea; actually, why are we using m4 here and not simple cat 2004-09-27 Michael Daum * utils/Makefile (1.10): - added testtcldocu to the default target 2004-09-27 Michael Daum * test/dbtest.sh (1.4), test/frobbingtest.pl.sh (1.9), test/glstest.pl.sh (1.9), test/itterlexicon.pl.sh (1.10), test/itterlexicon.tcl.sh (1.5), test/loadperlcdg.pl.sh (1.7), test/loadtclbindings.sh (1.8), test/netsearchtest.pl.sh (1.9), test/pointertest.tcl.sh (1.4), utils/testtcldocu.sh (1.2): - adopting new make system variables 2004-09-27 Michael Daum * test/cdgdbtest.c (1.4): - configure.h is config.h now 2004-09-27 Michael Daum * libcdg/Makefile (1.53): - removed unfinished parser modules from distribution 2004-09-27 Michael Daum * libcdg/cdg.c (1.87), libcdg/cdg.h (1.65), libcdg/cdgdb.c (1.21), libcdg/chart.h (1.3), libcdg/command.c (1.308), libcdg/compile.c (1.86), libcdg/constraintnet.c (1.177), libcdg/eval.c (1.112), libcdg/frobbing.c (1.231), libcdg/functions.c (1.50), libcdg/gls.c (1.85), libcdg/hook.c (1.38), libcdg/increment.c (1.72), libcdg/incrementalcompletion.c (1.112), libcdg/input.c (1.305), libcdg/levelvalue.c (1.69), libcdg/lexemgraph.c (1.140), libcdg/netsearch.c (1.117), libcdg/parse.c (1.190), libcdg/parse.h (1.63), libcdg/parser.y (1.87), libcdg/predicates.c (1.69), libcdg/scache.c (1.29), libcdg/scanner.l.m4 (1.28), libcdg/scorematrix.c (1.20), libcdg/set.c (1.19), libcdg/shift.h (1.5), libcdg/skel.h (1.11), libcdg/statistics.c (1.27), libcdg/testing.c (1.42), libcdg/timer.c (1.28), libcdg/timer.h (1.19), libcdg/transform.h (1.20), libcdg/variables.c (1.18), libcdg/write.c (1.65): - configure.h is config.h now - not using my_includeds any more 2004-09-27 Michael Daum * utils/depsy.pl (1.9): - fixed pod docu error 2004-09-27 Michael Daum * utils/cdg (1.25): - there's no funky .libs directory any more 2004-09-27 Michael Daum * grammar/penn/make-grammar.pl (1.2): - adopting new make system variables 2004-09-27 Michael Daum * grammar/negra/negra.m4 (1.24): - removed double pragmas - including base.cdg now, was base.m4 before 2004-09-27 Michael Daum * grammar/deutsch/make-adjectives.pl (1.32), grammar/deutsch/make-names.pl (1.17), grammar/deutsch/make-nouns.pl (1.25), grammar/deutsch/make-verbs.pl (1.46), grammar/negra/auto2silver.pl (1.3), grammar/negra/compare.sh (1.16), grammar/negra/find-unknown-names.pl (1.2): - adoption to make system variables 2004-09-27 Michael Daum * grammar/deutsch/doc.tex (1.55): - using times fonts now to be able to make nice pdfs now 2004-09-27 Michael Daum * grammar/deutsch/base.m4 (1.22): - added GPL header 2004-09-27 Michael Daum * doc/blah.conf.in (1.3), doc/libcdg.conf.in (1.14), doc/xcdg.conf.in (1.7), doc/yada.conf.in (1.7): - adoption to new file make system variables 2004-09-27 Michael Daum * bindings/hooker.c (1.13): - configure.h changed to config.h 2004-09-27 Michael Daum * cdgp/cdgp.c (1.47): - not using the my_includes any more as long as we mustn't 2004-09-27 Michael Daum * cdgp/BUILD (1.2): - removed useless BUILD indicator, moved it to the topdir 2004-09-27 Michael Daum * bindings/CDG/Makefile (1.1), bindings/CDG/Makefile.am (1.4), bindings/Makefile (1.1), bindings/Makefile.am (1.26), cdgp/Makefile (1.26), cdgp/Makefile.am (1.24), doc/Makefile (1.11), doc/Makefile.am (1.18), grammar/berlin/Makefile (1.3), grammar/berlin/Makefile.am (1.8), grammar/deutsch/Makefile (1.1), grammar/deutsch/Makefile.am (1.25), grammar/negra/Makefile (1.1), grammar/negra/Makefile.am (1.61), grammar/Makefile (1.1), grammar/penn/Makefile (1.1), grammar/penn/Makefile.am (1.16), grammar/stellingen/Makefile (1.29), grammar/stellingen/Makefile.am (1.11), grammar/Makefile.am (1.14), include/Makefile (1.1), include/Makefile.am (1.5), include/config.h (1.1), include/config.mk (1.1), include/configure.h (1.4), libcdg/Makefile (1.52), libcdg/Makefile.am (1.31), libcontainer/Makefile (1.29), libcontainer/Makefile.am (1.8), test/Makefile (1.16), test/Makefile.am (1.25), utils/Makefile (1.9), utils/Makefile.am (1.27), xcdg/Makefile (1.1), xcdg/Makefile.am (1.19), yada/images/Makefile (1.1), yada/images/Makefile.am (1.6), Makefile (1.70), Makefile.am (1.17), acinclude.m4 (1.34), bootstrap (1.8), configure (1.1), configure.in (1.32), yada/Makefile (1.1), yada/Makefile.am (1.16): - rooted out autotools due to extending frustrations: the new configure and make systems should be much cleaner and faster 2004-09-27 Kilian A. Foth * xcdg/parsetree.tcl (1.131): - stop nodeOfTimepoint from choking on -1 in completely cyclical trees - do not react to Control-Keyrelease events if they are part of something other than label editing 2004-09-27 Kilian A. Foth * xcdg/databrowser.tcl (1.12), xcdg/mytable.tcl (1.17): Give [Up], [Down], [PgUp] and [PgDn] the expected bindings in data browser widgets. 2004-09-27 Kilian A. Foth * utils/ccc-Makefile (1.11): Use make's -include feature instead of emulating it. 2004-09-24 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.90), grammar/deutsch/Grammatik.cdg (1.166), grammar/deutsch/Lexikon.cdg (1.128), grammar/deutsch/Namen.txt (1.71), grammar/deutsch/Nomen.txt (1.116), grammar/deutsch/TODO (1.9), grammar/deutsch/Verben.txt (1.128), grammar/deutsch/doc.tex (1.54), grammar/deutsch/extra.m4 (1.14): - random new lexicon items - allow NN to be ADV if it is reduplicated: `Faust auf Faust' - require GMOD to precede sibling PP: `der PIII meines Bosses mit 800 Mhz' *`der PIII mit 800 Mhz meines Bosses' - plug loopholes in ADV-Unterordnung: things like PTKSUP, PDS, PPER etc. should not be modified by ADV except in special cases - allow more varieties of reduced subclauses: `Dies ist ein erster Schritt, wenn auch nicht ausreichend.' `eine schnelle, wenn auch ungenaue Analyse' `Das ist ein schönes Ergebins, wenn auch kein Rekord.' 2004-09-24 Kilian A. Foth * utils/ccc-renumber (1.16): Give statistics about how many matches/mismatches there were. 2004-09-24 Kilian A. Foth * utils/tokenize.pl (1.41): Try to combine "Gates" + "'" into "Gates'" where appropriate. 2004-09-24 Kilian A. Foth * libcdg/frobbing.c (1.230), libcdg/transform.c (1.38), libcdg/transform.h (1.19): New frobbing method `threefold': do analysis of subproblems, then try to recombine the results unchanged, finally remove any remaining errors in the complete search space. This is the slowest but best method available, and is also how the results in our last publications were produced. The only difference is that now you get this automatically by just saying `frobbing'. To get what you used to get by saying just `frobbing', say `frobbing method=dyamic'. 2004-09-24 Kilian A. Foth * xcdg/parsetree.tcl (1.130): Mouse and keyboard selection of items can interfere when keyboard navigation is used while the mouse pointer happens to touch another widget. This is because some places assume that _activeItem == the item touched by the mouse == the one that the user wants to change. This is no longer true, so _activeItem is _mouseItem now, and labelSelect no longer implicitly assumes that it should change the global variable, but receives a parameter. 2004-09-24 Kilian A. Foth * xcdg/parsetree.tcl (1.129): Give {Alt,Control}+{Home,End} the obvious bindings. Fix word selection: self-modification prevention and wraparound can interact and must be checked together. When previewing a NIL attachment, choose the closest NIL node available. 2004-09-23 Michael Daum * test/hashtest.c (1.3): - fixed missing blahInitialization - fixed segfault when switching on verbosity 2004-09-23 Michael Daum * test/cdgtest.c (1.7): - added GPL header 2004-09-23 Michael Daum * libcontainer/string.c (1.11): - fixed segfault in strFinalize 2004-09-23 Michael Daum * libcontainer/hashtable.c (1.6), libcontainer/ringbuffer.c (1.5): - added missing include 2004-09-23 Michael Daum * libcontainer/blah.h (1.11): - fixed missing declaration of strInitialize and strFinalize 2004-09-23 Michael Daum * bindings/CDG/Utils.pm (1.9): - fixed pod docu 2004-09-23 Michael Daum * utils/cdg2latex.l (1.2): - fixed printf ing a % - fixed main() declaration showing up when switching all warnings and error messages 2004-09-23 Michael Daum * utils/cdgdbdump.c (1.5), libcontainer/bitstring.c (1.7): - added missing include 2004-09-23 Michael Daum * utils/indexer.c (1.10): - added missing include - removed unused variable 2004-09-23 Michael Daum * xcdg/parsetree.tcl (1.128): - fixed tcl docu 2004-09-23 Kilian A. Foth * libcdg/write.c (1.64): Guard against attribute/value identifiers that contain ' . 2004-09-22 Michael Daum * test/settest.c (1.1), test/stringtest.sh (1.7), test/vartest.c (1.1): - converted cdg demos to proper tests - added a tcl string tester 2004-09-22 Kilian A. Foth * xcdg/parsetree.tcl (1.127): Add mouseless editing to the parsetree widget. Born of the frustration of editing trees on a laptop with the pathetic number of two mouse buttons, this addition aims to provide the same functionality as the popular dragging and clicking commands. The key bindings reflect my personal preferences, and I have deliberately introduced eight handlers instead of one so you cannot easily cange them, bwuaahaahaahaaha . No, seriously, the reason for the many handlers is that I can't figure out how to transmit information about active modifier keys through Tcl's %x mechanism, so I needed one handler for Shift-ed keypresses, one for Control-ed etc. The actual bindings right now are: Left/Right/Up/Down select the word on which you want to operate. The selected word is highlighted as if it had been touched by the mouse. The Down action is, of course, ill-defined because many words have multiple dependents, and does nothing in this case. Shift-Left/Up/Right/Down change the attachment of that word. Control-Up/Down change the label of that word. Alt-Up/Down change the lexical reading of that word. As a bonus, Enter in combination with any of these modifiers emulates a right click, i.e. it optimizes the attachment/label/lexical reading unarily. (An alternative to this solution would be simply to reuse the existing menu for labels or lexicon items, but then one would need to change it so it provides for autocorrection as well - the entire mouseless editing idea isn't worth much without that.) 2004-09-22 Kilian A. Foth * utils/ccc-parse (1.12), utils/ccc-distribute (1.1): After parsing automatically, distribute the results into subdirectories 001/ etc., automatically, no matter whether there are many or few of them. 2004-09-21 Michael Daum * acinclude.m4 (1.33): improved swig detection 2004-09-21 Kilian A. Foth * xcdg/parsetree.tcl (1.126): Remove unused findLabel (duplicates labelOfEdge). Introduce similar edgeOfTimepoint and wordOfTimepoint and actually use them everywhere. 2004-09-20 Kilian A. Foth * utils/deutsch-tagger.pl (1.16): Use papa.lex only if present (trying to use it if it is not present would cause TnT to exit). 2004-09-20 Kilian A. Foth * utils/ccc-parse (1.11): With -f, actually use the previously found analysis instead of just glancing at it uncomprehendingly. 2004-09-13 Kilian A. Foth * xcdg/parsetree.tcl (1.125): - remove debug puts - restore variable constraint highlighting 2004-09-10 Kilian A. Foth * xcdg/parsetree.tcl (1.124): Make arcs sensitive to right clicks. 2004-09-10 Kilian A. Foth * xcdg/parsetree.tcl (1.123): Fix edgeRightClick{}: edge right click retrieved the old regent incorrectly, which totally broke undo after right click. 2004-09-10 Kilian A. Foth * libcdg/tagger.c (1.60): Plug file handle leak introduced in 1.59. 2004-09-09 Othello Maurer * libcdg/command.c (1.307): quote characters on command line you can quote characters in strings on the command line with a backslash now for example: lexicon '\'' will present you the lexicon entry for a single ' 2004-09-08 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.89), grammar/deutsch/Grammatik.cdg (1.165), grammar/deutsch/Hierarchien.cdg (1.47), grammar/deutsch/Lexikon.cdg (1.127), grammar/deutsch/Namen.txt (1.70), grammar/deutsch/Nomen.txt (1.115), grammar/deutsch/TODO (1.8), grammar/deutsch/Verben.txt (1.127), grammar/deutsch/doc.tex (1.53): - random new lexicon items - detect misleading `um' - distinguish crossing of `:': from the right it is not so bad, from the left REALLY bad. - lexicalize adverb/pronoun subordinations: praktisch keine *heute keine viel mehr *insgesamt meine - check case agreement across three edges for `eine Million Aktien verkaufen' - constrain order of OBJP and PP: `Wir gehen am Sonntag in die Kirche.' *`Wir gehen in die Kirche am Sonntag.' - punish [cat:APPR]--PN-->[] edges extra hard for their lengths - plug hole in detection of impersonal passive (`sein' can form passives!) - ADJA can carry APP modifications. - ADJA can be NP2. (Exceptions for ADJA are now in all places where NP are expected. I suppose I should have made `Nominal' subsume ADJA in the first place.) - cover `an die 100 Sorten' - cover adverbial FM: `Last/ADV not least ist auch der Umsatz zurückgegangen.' 2004-09-06 Michael Daum * yada/Makefile.am (1.15): - added test to check wellformedness of tcl documentation 2004-09-06 Michael Daum * xcdg/allhierarchies.tcl (1.33), xcdg/alllevels.tcl (1.36), xcdg/alllexemes.tcl (1.30), xcdg/allparses.tcl (1.61), xcdg/allwordgraphs.tcl (1.68), xcdg/balloon.tcl (1.8), xcdg/busy.tcl (1.23), xcdg/commandhistory.tcl (1.3), xcdg/commands.tcl (1.117), xcdg/parsetree.tcl (1.122), xcdg/compat.tcl (1.5), xcdg/databrowser.tcl (1.11), xcdg/parse.tcl (1.77), xcdg/shell.tcl (1.59), xcdg/textutils.tcl (1.8), xcdg/visparses.tcl (1.50), yada/YadaConfigItem.tcl (1.7), yada/YadaGenerals.tcl (1.14), yada/YadaMain.tcl (1.18), yada/YadaProc.tcl (1.21), yada/YadaRanking.tcl (1.18), yada/YadaRunner.tcl (1.27): - fixed tcl documentation reported by "make check" 2004-09-06 Michael Daum * xcdg/Makefile.am (1.18): - added a test to check wellformedness of tcl documentation (try make check) 2004-09-06 Michael Daum * doc/tclfilter.pl (1.6): - added some error reports on malformed tcl documentation - fixed derivation of memberTypes CAUTION: right now class members of the same name clash, e.g. a variable confirmexit and a methoid confirmexit - besides of it being a bad habit (tm). 2004-09-06 Michael Daum * utils/testtcldocu.sh (1.1): - added tool to test correctness of tcl source documentation 2004-09-06 Kilian A. Foth * xcdg/parsetree.tcl (1.121): When dragging an edge around, highlight not only the nearest node but also its corresponding word, so you can see better where it would snap to if dropped. 2004-09-06 Kilian A. Foth * libcdg/input.c (1.304): Update `phrase quotes' counter correctly when sentences start with ". 2004-09-03 Michael Daum * debian/changelog (1.2), debian/control (1.2), debian/dirs (1.2), debian/rules (1.2): ... this is all so new 2004-09-03 Michael Daum * grammar/deutsch/Makefile.am (1.24): - cleaning backup files "*~" 2004-09-03 Michael Daum * grammar/berlin/Makefile.am (1.7): - fixed grammar compilation 2004-09-03 Michael Daum * configure.in (1.31): - keep the berlin makefile in the boat; could not build it before 2004-09-03 Michael Daum * acinclude.m4 (1.32): - removed double check - fixed progress comments 2004-09-03 Michael Daum * Makefile.am (1.16): - clean backup files "*~" - fixed deb packageing - added target ChangeLog 2004-09-03 Michael Daum * AUTHORS (1.4): - collection of email addresses in the format that cvs2cl needs it 2004-09-03 Michael Daum * libcontainer/Makefile.am (1.7), test/Makefile.am (1.24), utils/Makefile.am (1.26), xcdg/Makefile.am (1.17), yada/images/Makefile.am (1.5), yada/Makefile.am (1.14): - clean backup files "*~" 2004-09-03 Michael Daum * bindings/CDG/Makefile.am (1.3), bindings/Makefile.am (1.25), cdgp/Makefile.am (1.23), doc/Makefile.am (1.17), grammar/negra/Makefile.am (1.60), grammar/penn/Makefile.am (1.15), grammar/stellingen/Makefile.am (1.10), grammar/Makefile.am (1.13), include/Makefile.am (1.4), libcdg/Makefile.am (1.30): - clean backup files "*~" 2004-09-03 Michael Daum * utils/cdg (1.24): - fixed variable typo 2004-09-03 Kilian A. Foth * utils/tokenize.pl (1.40): Option `-h' explains usage. Option `-f' gives flat output (one word per line). 2004-09-03 Kilian A. Foth * utils/ccc-todo (1.2): List files in the right order, not alphabetically. 2004-09-02 Michael Daum * ChangeLog (1.14): This will be generated on demand from the cvs log messages. 2004-09-02 Kilian A. Foth * libcdg/parse.c (1.189): parseOptimizeStructure(): Do not allow result == wordno, since that would violate the built-in, implicit, secret, never-openly-spoken-of self-modification constraint. 2004-09-02 Kilian A. Foth * acinclude.m4 (1.31): Simplify SuSE check. 2004-09-02 Kilian A. Foth * utils/ccc-search (1.10): Fix logic error in count_statistics(). 2004-09-02 Kilian A. Foth * acinclude.m4 (1.30), configure.in (1.30): Check for recent version of Flex. Check for versions of BDB that are not called libdb.so. 2004-09-01 Kilian A. Foth * xcdg/parsetree.tcl (1.120): Oops, edge optimization by edge right click overwrites the old binding of label menu by edge right click (needed when you can't touch the current label because it is the empty string). In the interest of consistency, edge right click remains at "optimize direction" and the much more rarely needed "pop up ancillary label menu" is on edge middle click now. 2004-09-01 Kilian A. Foth * libcdg/parse.c (1.188): parseOptimizeStructure(): Consider NIL as well. 2004-09-01 Lidia Khmylko * grammar/deutsch/Verben.txt (1.126): valencies fixed 2004-09-01 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.164): errors fixed 2004-09-01 Lidia Khmylko * grammar/deutsch/Adjektive.txt (1.88): adjective removed 2004-09-01 Michael Daum * utils/Makefile.am (1.25): - fixed "make install" (see log messages on cdg) - the files cdg, text2png.pl and ccc-Makefile are patched beforehand to fix the default values in them - ccc-Makefile was installed as a shell script but is only yet another data file - all scipts are installed to $prefix/bin now; was $prefix/share/cdg/utils before plus sym links to from $prefix/bin to $prefix/share/cdg/utils; therefor the symlink thingy is gone; was busted anyway - added uninstall target to remove the xcdg link 2004-09-01 Michael Daum * utils/cdg (1.23): - added new functions try_bindings and try_xcdg following the example of try_binary - added variables * CDGSYSTEM: points to the system installation * CCC_DIR: points to the corpus collection This allows to have cdg installed systemwide and keep a per user CDGHOME sandbox with his private grammars and scripts etc. - during "make install" the variables CDGSYSTEM and CCC_DIR are patched to match the actual installation paths 2004-09-01 Kilian A. Foth * xcdg/parse.tcl (1.76), xcdg/parsetree.tcl (1.119), libcdg/parse.c (1.187), libcdg/parse.h (1.62): parseOptimizeStructure() does the same as parseOptimizeLabel(), only for subordination, and edge right clicks now exploit this. Very useful for rearranging words along a long auxiliary chain (which is usually scrolled off the screen). 2004-09-01 Michael Daum * utils/cdg2png.pl (1.3): - fixed "docu" 2004-09-01 Michael Daum * utils/indexer.c (1.9): - added GPL header (need to do this once and for every file in cdg) 2004-09-01 Michael Daum * libcdg/timer.h (1.18): - fixed doxygen warnings 2004-09-01 Michael Daum * libcdg/scorematrix.c (1.19), libcdg/scorematrix.h (1.7): - removed redundant function docu from headerfile - fixed doxygen warnings 2004-09-01 Michael Daum * libcdg/parse.c (1.186): - fixed doxygen docu 2004-09-01 Michael Daum * libcdg/lexemgraph.c (1.139), libcdg/lexemgraph.h (1.68): - doxygenized the lexemgraph module - removed unused function lgNodeFollowsDirectly() - moved inputSpuriousUppercase() in here and renamed it to lgSpuriousUppercase(); this function was solely used in the lexemgraph module anyway and by the way fixes a cycle c-header dependency 2004-09-01 Michael Daum * libcdg/input.c (1.303), libcdg/input.h (1.141): - reworded typedefinitions recursive structures and using global forward definitions from cdg.h now, we don't need any struct MyTypeStruct *foo anymore as the MyType is always defined in advance. Casting in unexpected places is gone aswell. - removed unused function levelFollowsDirectly() - moved cvAnalyse to the constraintnet module - moved inputSpuriousUppercase(LexemGraph, Arc) to the lexemgraph module and renamed it tp lgSpuriousUppercase(LexemGraph, Arc) - fixed doxygen warnings 2004-09-01 Michael Daum * libcdg/eval.c (1.111), libcdg/eval.h (1.44): - removed non-matching prototype argument namings (this is deprecated anyway) - added doxygen docu for the static_strings - fixed doxygen docu of evalCurrentFormula - fixed doxygen docu of evalTerm (uhh this never was right) 2004-09-01 Michael Daum * libcdg/constraintnet.c (1.176), libcdg/constraintnet.h (1.65): - removed unused function cnNodeFollowsDirectly - added doxygen docu for cnUseNonSpec - moved cvAnalyse() from the input module in here 2004-09-01 Michael Daum * libcdg/chunker.h (1.12): - moved forward declarations to cdg.h and fixed ChunksStruct definition to play well with the forward declarations 2004-09-01 Michael Daum * libcdg/chunker.c (1.41): - fixed header inclusion - fixed doxygen docu 2004-09-01 Michael Daum * libcdg/cdgdb.h (1.10): - fixed header inclusion - removed argument names; they are never needed and rendundant and if they don't match they in hit doxygen 2004-09-01 Michael Daum * libcdg/cdgdb.c (1.20): - fixed doxygen comments 2004-09-01 Michael Daum * libcdg/cdg.c (1.86), libcdg/cdg.h (1.64): - moved forward type definitions to cdg.h to get cyclic C-structures done right; before that, several headers did their own redundant forward definitions; So new policy: put them all into cdg.h and #include it. 2004-09-01 Michael Daum * configure.in (1.29), acinclude.m4 (1.29): - added ccc support: ./configure --with-ccc-dir=DIR specifies the directory where the corpus collection is. defaults to /data/cvs/ccc 2004-09-01 Michael Daum * bindings/Makefile.am (1.24): - reverted previous cvs commit: don't make things worse, please 2004-09-01 Michael Daum * bindings/hooker.h (1.7): - cleaned up doxygen comments 2004-09-01 Michael Daum * doc/libcdg.conf.in (1.13), doc/xcdg.conf.in (1.6), doc/yada.conf.in (1.6): - updated doxygen config file to 1.3.8 - disabled buggy JAVADOC_AUTOBRIEF option (bug report already filed to the developers of doxygen) 2004-09-01 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.45): Fix c&p error that prevented the constraint AVZ-OBJP-loctype from working. 2004-09-01 Kilian A. Foth * utils/ccc-search (1.9): Fix output at verbosity 0: The order of edges within the tree that `showparse' needs is more complicated than just s->from, since there are multiple levels. AnnoEntry2hash used to calculate this while translating, but since it isn't used anymore we have to re-calculate this when we need it. Think of it as the moral equivalent to parseIndex() for Parses. 2004-09-01 Michael Daum * doc/libcdg.tex (1.92): - doxygenized docu of the lexemgraph module 2004-09-01 Michael Daum * doc/Makefile.am (1.16): - patching automatically generated tex docu isn't needed any more with newer doxygen versions 2004-08-31 Michael Daum * test/vectortest.c (1.6): - fixed comments 2004-08-30 Michael Daum * yada/Process.tcl (1.15): - fixed commandline of the process invocation 2004-08-30 Michael Daum * README (1.6): - clarified the newest flex changes: bye bye flex 2.5.4a 2004-08-30 Michael Daum * yada/YadaMachine.tcl (1.12): - removed debug messages - make fping quiet 2004-08-30 Michael Daum * yada/YadaGrammar.tcl (1.23): - a process' pid is always > zero 2004-08-30 Michael Daum * yada/YadaExperiment.tcl (1.10): - removed debug message 2004-08-30 Michael Daum * xcdg/visparses.tcl (1.49): - removed uggly static fonts 2004-08-30 Michael Daum * utils/cdgdbdump.c (1.4), utils/indexer.c (1.8): - applied Yannicks db patch to work with newer versions (>4.0) 2004-08-30 Michael Daum * utils/cdg (1.22): - try static binaries _after_ normal once 2004-08-30 Michael Daum * cdgp/cdgp.c (1.46): - the signal handlers are initialized before running the init files again as the new flex versions don't segfault now getting an SIGINT. - added gpl header to our main(:-) 2004-08-30 Michael Daum * libcdg/cdgdb.c (1.19), acinclude.m4 (1.28): - applied patch from Yannick to deal with interface changes in db versions > 4.0 2004-08-30 Michael Daum * libcdg/constraintnet.c (1.175): - fixed comments 2004-08-30 Michael Daum * libcdg/eval.c (1.110): - clarified comment 2004-08-30 Michael Daum * libcdg/input.c (1.302), libcdg/input.h (1.140): - new function formatTime to get a human readable string from milliseconds 2004-08-30 Michael Daum * libcdg/scanner.l.m4 (1.27): - fix for new flex interface: no more dependency on old and buggy flex version; no more segfault on ctrl-c 2004-08-30 Michael Daum * libcdg/Makefile.am (1.29): - fixed scanner dependencies - fixed CTAGS generation - removed tags target from the all target 2004-08-30 Kilian A. Foth * utils/ccc-search (1.8): eval_op(): Correctly handle the pseudofeature `no'. 2004-08-26 Bjoern Engelmann * xcdg/allhierarchies.tcl (1.32): feature added: the hierarchy-highlighting is now bidirectional various minor bugfixes: mainly performance issues 2004-08-20 Kilian A. Foth * utils/ccc-annotate (1.2): Remove debug output. 2004-08-20 Kilian A. Foth * utils/ccc-annotate (1.1): This script automatically finds the next sentence that an annotator should read and opens it in XCDG. They can then simply use the `save' and `next' command in the tree editor to advance through the corpus (provided that the missing hand annotations are consecutive). 2004-08-19 Bjoern Engelmann * xcdg/allhierarchies.tcl (1.31): bugfix: deutsch.cdg can now be viewed in the hierarchy tab 2004-08-18 Bjoern Engelmann * xcdg/allwordgraphs.tcl (1.67): removed the details-button which is useless. removed edit- and delete-button which were invisible anyway. 2004-08-17 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.163): a rule improved 2004-08-17 Kilian A. Foth * utils/ccc-search (1.7): Add missing parameter to edge_length() calls. 2004-08-17 Kilian A. Foth * utils/search-annotations.pl (1.44): search-annotations.pl is now called ccc-search. Use the faster new version. 2004-08-16 Kilian A. Foth * utils/ccc-search (1.6): - re-enable sorting the list of all trees (the numbers serve as an informal progress indicator while searching) - remove evil global variable $current_AnnoEntry, $a is passed through the call chain now. - take care to reliably exit all loops after interruption 2004-08-16 Kilian A. Foth * utils/ccc-search (1.5): - remove the `lookup' pseudo-function; it was already broken before the rewrite, and it encourages sloppy annotating. If you want to analyse tense distribution, you should, by God, write tense information into you bloody trees and not try to re-deduce them after the fact. - fix error within eval_path(): The `^' branch carefully retrieved the alternate AnnotationStruct $a and then used $anno anyway - fix error while calling eval_path(): arrays do not interpolate into the call stack, so you must call a function with prototype ($$$$) as func($$a[0], $$a[1], $x, $y) rather than func($a, $x, $y) even though @a has to elements. 2004-08-16 Michael Daum * utils/depsy.pl (1.8): - better warning about not having dependency labels 2004-08-16 Kilian A. Foth * utils/cdg (1.21): Try harder to find the actual program, so that it also works in the installed tree. 2004-08-16 Kilian A. Foth * utils/Makefile.am (1.24): Ensure that all important utilities are available under $PREFIX/bin. 2004-08-16 Kilian A. Foth * bindings/Makefile.am (1.23): Fudge an extra symbolic link so that an installed xcdg finds its .so files. 2004-08-16 Kilian A. Foth * configure.in (1.28): Change xcdgdir so that an installed xcdg actually works. 2004-08-13 Kilian A. Foth * utils/Makefile.am (1.23): Remove outdated source file reference. 2004-08-13 Kilian A. Foth * doc/doc.tex (1.85): Document predicate `height'. 2004-08-13 Kilian A. Foth * grammar/deutsch/Adjektiv-Templates.txt (1.3), grammar/deutsch/Adjektive.txt (1.87), grammar/deutsch/Grammatik.cdg (1.162), grammar/deutsch/Hierarchien.cdg (1.46), grammar/deutsch/Lexikon.cdg (1.126), grammar/deutsch/Namen.txt (1.69), grammar/deutsch/Nomen.txt (1.114), grammar/deutsch/Verben.txt (1.125), grammar/deutsch/base.m4 (1.21), grammar/deutsch/doc.tex (1.52), grammar/deutsch/makros.m4 (1.14): - random new lexicon items - sort all shallow parsing constraints into section `shallow' - give some guidelines for naming constraints - use rightpenalty and leftpenalty only for noun modification - modification of verbs has to follow the stricter topology constraints that would override these preferences anyway. - detect and signal sentences that should show inversion but don't - forbid left modification of PWS - remove redundant flexion class constraints - move exists() into the premise in all agreement rules - pair degrees and comparators correctly: positive/wie, comparative/als - disprefer the category mismatch `Halt, oder ich schieße!' - allow PDS to carry OBJC since it might refer to a noun that can carry OBJC - restrict [infinitive:bleiben]+[cat:VVINF] to full verbs (further restrictions apply that I don't quite grasp) - OBJP should appear closer to the verb than PP - disambiguate PRED vs. SUBJ in subclauses - punish long REF edges with a distance constraint rather than with multiple structural constraints - enforce low attachment of REL to APP chains - disprefer all kinds of metagrammatical usage - always record verb tense and mood, whether redundant or not 2004-08-13 Kilian A. Foth * utils/ccc-search (1.4): sub has(): add forgotten `= @_' 2004-08-13 Kilian A. Foth * utils/ccc-search (1.3): Delete the temporary input file immediately after loading it (instead of at the end), so they won't pile up if you debug a crashing version. 2004-08-13 Kilian A. Foth * utils/ccc-search (1.2): Fix mistyped `att3'--> `att2'. 2004-08-13 Kilian A. Foth * utils/ccc-search (1.1), utils/ccc-search.pl (1.5): Remove filename extension. Replace SpecificationStruct_name_set, which does the wrong thing, with new inputSpecSetName(), which understands our String type. 2004-08-13 Kilian A. Foth * libcdg/input.c (1.301), libcdg/input.h (1.139): Provide accessor function for the ->name field of a Specification. (Swig's auto-generated _set() function does the wrong thing, it modifies the shared string.) 2004-08-12 Othello Maurer * utils/ccc-search.pl (1.4): [no log message] 2004-08-12 Othello Maurer * libcdg/input.h (1.138): made AnnotationStruct and SpecificationStruct available for SWIG 2004-08-12 Othello Maurer * utils/ccc-search.pl (1.3): [no log message] 2004-08-12 Othello Maurer * utils/ccc-search.pl (1.2): fixed some bugs 2004-08-12 Othello Maurer * bindings/cdg.i (1.15): added freeSpec() for deleting a SpecificationStruct 2004-08-12 Othello Maurer * utils/ccc-search.pl (1.1): ccc-search pl is a new version of search-annotations.pl. it uses the original c structures of annotations for processing instead of parsing them into new perl hash structures. thus memory usage can be minimized. 2004-08-12 Kilian A. Foth * utils/tokenize.pl (1.39): Keep track of parenthesis level even in the face of words with multiple parentheses in them. 2004-08-12 Kilian A. Foth * utils/deutsch-tagger.pl (1.15): Detect some PIS mistagged as ADV. 2004-08-12 Kilian A. Foth * utils/ccc-renumber (1.15): Increase max_drift to the values actually needed for heiseticker. Move results from NEW to their place more intelligently. 2004-08-11 Michael Daum * grammar/penn/pennplugin.pl (1.9): complete rework: - work on reasonable dependency edge labels - added pennWriteLexiconEntry callback: exclude hand written lexincon entries - added pennReadSentence callback: fix obvious tagging errors - adde penPostProcess callback: warn about unset dependency edge labels - added appoFixHandler: distinguish appos and compounds - added a posHandler: a posessive nominal attachment is an ADJ - parenthesisHandler: By default, a parenthesis is attached to the head of the matrix constituency. This shifts it to the first element just before the parenthesis. - fixed setEdgeTag handler: warn on a real overwrite - added compoundHandler: bundle well known compounds as such and separate them from the rest of the constituency; this needs a list of well known compounds; there's NO (zero) way to get NPs reasonably extacted from wsj w/o such a list; see the examples in the comments - added coordinationHandler build our model of a coordination as a left attached dependency chain - added the npCoordinationHandler: This is a special version of the general coordinationHandler needed because coordinations inside an NP are done differently as in the general case - if at all, i.e. this is a very common example of flattening an NP with no markup what's happening: "the new sales and marketing arm" ... - added puncHandler: detach punctation - added numberHandler: bundles adjacent cardinals as one cardinal phrase - added sbarFixHanbdler: unify annotation guide, e.g. in s79, s1458, s2136 s5042, s5401 and the like CAUTION: this version generates lots of warnings about unset dependency edges 2004-08-11 Michael Daum * grammar/penn/penn.m4 (1.7): - added auto-lexicon.cdg and auto-grammar.cdg - more work on a reasonable distance based penalty formular 2004-08-11 Michael Daum * grammar/penn/lexicon.cdg (1.3): - added determiners, pronouns 2004-08-11 Michael Daum * grammar/penn/grammar.cdg (1.4): - defined SYN level and a reasonable set of dependency edge labels (see the comments) 2004-08-11 Michael Daum * grammar/penn/Makefile.am (1.14): - added auto-grammar.cdg target - generating a toy lexicon to allow xcdg parse evaluation - fixed clean target to erase stage cdas and pngs - fixed penn.cdg target to use penn.m4 - by default only the first 140 sentences are converted: please adjust the -max 140 entry to your needs 2004-08-11 Michael Daum * grammar/penn/test-transform-sentence.pl (1.1): - taken from negra and adapted to penn treebank 2004-08-11 Michael Daum * grammar/penn/make-grammar.pl (1.1): - generates a toy grammar that generates unary label definition constraints 2004-08-09 Michael Daum * libcdg/compile.c (1.85): - removed debug flags from grammar cflags, increase gcc optimization level 2004-08-09 Michael Daum * libcdg/lexemgraph.c (1.138): - fixed memory leak 2004-08-09 Michael Daum * libcdg/compile.c (1.84), libcdg/functions.c (1.49), libcdg/functions.h (1.23): with Kilian - added translations for parents() and height() - added pointer to vs accu for comTranslateLexemnodeAccess as lexemes might not only com from lvs but also from the context 2004-08-06 Kilian A. Foth * utils/eval-tagger.pl (1.14): Fix mistyped TreeTagger invocation. 2004-08-05 Kilian A. Foth * libcdg/functions.c (1.48), libcdg/functions.h (1.22): Function `height' returns the number of levels in the highest treee below a word. Not actually used, but something that one might very well want. 2004-08-05 Kilian A. Foth * utils/ccc-check (1.16): Process subdirectory `001' if present. 2004-08-02 Kilian A. Foth * utils/ccc-renumber (1.14): Word mismatches weren't detected when the only change was that `New' `York' was changed to `New York'. Now we use # instead of space as a marker in the comparison strings. 2004-07-29 eugen * Makefile.am (1.15): added new deb target: produces debian binary package 2004-07-29 eugen * debian/dirs (1.1), debian/files (1.1), debian/rules (1.1): new debian support 2004-07-29 eugen * debian/control (1.1), debian/copyright (1.1): new debian support 2004-07-29 eugen * debian/README.debian (1.1), debian/changelog (1.1): new debian package support 2004-07-29 Kilian A. Foth * xcdg/parsetree.tcl (1.118): wordEnter{}: do not access LexiconItems in abtract parses (was broken in yesterday's code shuffling) 2004-07-28 Kilian A. Foth * libcdg/parse.c (1.185), libcdg/parse.h (1.61), xcdg/parse.tcl (1.75), xcdg/parsetree.tcl (1.117): Unify tool tips in the parse editor: middle click on a conflict *or* a word will pop up the full definition. Touching a word will only pop up the ->description. This makes the parse window much less busy and more consistent. 2004-07-27 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.161): spelling 2004-07-27 Lidia Khmylko * grammar/deutsch/Adjektive.txt (1.86), grammar/deutsch/Namen.txt (1.68), grammar/deutsch/Nomen.txt (1.113): lexics 2004-07-26 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.44): - add special provision for `OBJC' etc. to modify a PDS - allow ss/ß alternation even for participial adjectives 2004-07-22 Lidia Khmylko * grammar/deutsch/Namen.txt (1.67), grammar/deutsch/Nomen.txt (1.112): lexics added 2004-07-22 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.160): examples improved 2004-07-22 Kilian A. Foth * grammar/negra/negraplugin.pl (1.74): Normalize relative clauses to modify the last noun of multiword-NP. 2004-07-22 Kilian A. Foth * utils/shift-reduce.pl (1.14): - parse all available .cda files if pointed to a directory - new option -x immediately launches the tree viewer 2004-07-22 Kilian A. Foth * utils/eval-tagger.pl (1.13): - deal with non-quoted category specifications - print sentence id when reporting errors rather than just `WRONG!' 2004-07-22 Kilian A. Foth * utils/tokenize.pl (1.38): - special token `\nobreak' may be used to mark inter-token gaps that are not to be considered sentence endings - recognise some names that end in ! such as `Yahoo!'; these also do not end sentences. The approach to do this is of course wrong; lists of such perverted names should come in as a user option and not be hardcoded. 2004-07-22 Kilian A. Foth * utils/ccc-renumber (1.13): Insert fake file modification times so that auto-renaming does not count as a `change' for backup tools. This is arguably a lie, but much easier for me personally; rather than sync thousands of trees between my home directories whenever a corpus changes, it is much easier to sync just the .txt and re-run the renumbering command afterwards. 2004-07-22 Kilian A. Foth * utils/ccc-rename (1.3): New option -l hacks the lattice name as well as the annotation name. 2004-07-22 Kilian A. Foth * xcdg/parsetree.tcl (1.116): drawAll{}: do not attempt to access non-existing itk_component(conflicts). 2004-07-22 Kilian A. Foth * xcdg/parsetree.tcl (1.115): In abstract parses the word menu would be empty. Do not pop it up. 2004-07-22 Kilian A. Foth * utils/shift-reduce.pl (1.13): When reading a transition table, re-create option settings that were in effect when the table was created. 2004-07-21 Kilian A. Foth * utils/deutsch-tagger.pl (1.14): Detect mistagged `am meisten'. 2004-07-16 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.159): examles and rules are imroved; I have marked my questions with the word 'achtung'. 2004-07-16 Kilian A. Foth * libcdg/input.c (1.300): inputLexiconQuery() must also probe for words that could be generated from templates, otherwise the evaluation of several parses might change depending on the order they were checked. 2004-07-16 Kilian A. Foth * xcdg/parsetree.tcl (1.114): Correctly position the word_description balloon. 2004-07-16 Kilian A. Foth * utils/tokenize.pl (1.37): Tokenize `(un)heimlich' into just one word. 2004-07-16 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.85), grammar/deutsch/Grammatik.cdg (1.158), grammar/deutsch/Hierarchien.cdg (1.45), grammar/deutsch/Lexikon.cdg (1.125), grammar/deutsch/Nomen.txt (1.111), grammar/deutsch/Verben.txt (1.124), grammar/negra/known_errors (1.46): - random new lexicon items - use `Inversion' instead of `scrambling' for re-ordered subclause VPs (`scrambling' already means something else). - forbid adverbials after attributes - detect misordered predicatives in subclauses - require comma after appositions - require comma for `aber' etc. 2004-07-15 Michael Daum * bindings/hooker.c (1.12): - simplified logPrintf 2004-07-15 Michael Daum * bindings/hooker.c (1.11): - removed possible buffer overflow in logPrintf 2004-07-14 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.157): examples improved 2004-07-14 Kilian A. Foth * xcdg/parsetree.tcl (1.113): Remove debug puts. 2004-07-14 Kilian A. Foth * utils/ccc-todo (1.1): Script to find automatically created trees with no corresponding hand-approved tree. 2004-07-14 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.156): rules are imroved 2004-07-14 Michael Daum * libcdg/tagger.c (1.59): - fixed insecure creation of temp files 2004-07-14 Kilian A. Foth * grammar/deutsch/make-names.pl (1.16): Generate genitive forms for female first names. 2004-07-14 Kilian A. Foth * utils/deutsch-tagger.pl (1.13): Further rules about verb prefixes and reflexive/nonreflexive pronouns. 2004-07-14 Kilian A. Foth * utils/ccc-check (1.15): Guard against UNIX's dread `argument list too long' message. 2004-07-14 Kilian A. Foth * utils/tokenize.pl (1.36): Detect abbreviations like `z.B.' that really should be two words (as opposed to abbreviations like `Y.M.C.A.' which should really be one word). 2004-07-14 Kilian A. Foth * xcdg/balloon.tcl (1.7), xcdg/parsetree.tcl (1.112), xcdg/visparses.tcl (1.48), libcdg/input.c (1.299), libcdg/input.h (1.137): Middle click on a conflict will now pop up the constraint on the canvas. Much easier than switching to the constraint box, selecting and pressing `show'. (Lots of C code had to be changed because we had only printConstraint() and not formatConstraint().) 2004-07-14 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.84), grammar/deutsch/Grammatik.cdg (1.155), grammar/deutsch/Lexikon.cdg (1.124), grammar/deutsch/Namen.txt (1.66), grammar/deutsch/Nomen.txt (1.110), grammar/deutsch/Verben.txt (1.123), grammar/deutsch/doc.tex (1.51): - random new lexicon items - use `Inversion' instead of `scrambling' for re-ordered subclause VPs (`scrambling' already means something else). - forbid adverbials after attributes - detect misordered predicatives in subclauses - require comma after appositions - require comma for `aber' etc. 2004-07-14 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.43): Consolidate semi-specified morphology generation. 2004-07-14 Kilian A. Foth * utils/ccc-renumber (1.12): Option `-d' searches for trees only in specified dir. Old `debug' option is now `-t' (`test'). 2004-07-13 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.154): rules are improved 2004-07-13 Michael Daum * yada/CdgProcess.tcl (1.14), yada/yada.sh (1.14): - removed workarround left back 2004-07-12 Olga Gertchikova * grammar/deutsch/Lexikon.cdg (1.123), grammar/deutsch/Verben.txt (1.122): [no log message] 2004-07-12 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.153): rules are improved 2004-07-12 Lidia Khmylko * grammar/deutsch/Namen.txt (1.65): a first name added 2004-07-12 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.152): examples improved 2004-07-12 Kilian A. Foth * libcdg/parse.c (1.184): Fix obscure timing error: since word swaps can change the reading of a word, but the reading is not actually used to compute similarity by parseFindLn(), newln may actually be == oldln even if the reading does change. Therefore w->word must be changed before the self-assignment check. 2004-07-09 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.151): rules and examples are improved 2004-07-09 Olga Gertchikova * grammar/deutsch/Nomen.txt (1.109): [no log message] 2004-07-08 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.150), grammar/deutsch/Nomen.txt (1.108), grammar/deutsch/Verben.txt (1.121): [no log message] 2004-07-07 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.149): rules are improved 2004-07-07 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.148): examples improved 2004-07-06 Lidia Khmylko * grammar/deutsch/Nomen.txt (1.107): the feature "set" added 2004-07-06 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.147): examples improved 2004-07-06 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.146): rules are improved 2004-07-01 Othello Maurer * libcdg/compile.c (1.83): - Slightly changed _lookup. If the map could not be found, the function will return an empty string now instead of aborting with an 'arguments must not be null' error. - comTranslateLookup now generates different code which provides warnings on type mismatches. - comTranslateHas now behaves correctly when the 6th argument is -1 2004-07-01 Othello Maurer * libcdg/functions.c (1.47): - Slightly changed funcLookup. If the map could not be found, the function will return an empty string now instead of aborting. 2004-06-29 Olga Gertchikova * grammar/deutsch/Verben.txt (1.120): valency "a" added to the verb "enden" 2004-06-29 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.145): rules and examples improved 2004-06-29 Kilian A. Foth * utils/tokenize.pl (1.35): Silence warning about useless empty `push'. 2004-06-29 Kilian A. Foth * utils/ccc-Makefile (1.10): Delete intemediary .cdg files upon failed creation; having them existing but empty causes subtle errors because of changed lexical coverage. 2004-06-29 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.144), grammar/deutsch/Hierarchien.cdg (1.44): Correct some contraints and comments. 2004-06-29 Kilian A. Foth * grammar/deutsch/Nomen.txt (1.106), grammar/deutsch/Adjektive.txt (1.83), grammar/deutsch/Lexikon.cdg (1.122), grammar/deutsch/Verben.txt (1.119): Random new lexicon items. 2004-06-29 Kilian A. Foth * grammar/deutsch/base.m4 (1.20), grammar/deutsch/make-verbs.pl (1.42): Correct obscure lexicon lookup failure: if a separable prefix is homonymous with different kind of truncation defined as a template, it blocks the use of that template even under templates=bycategory, since the difference is in cat2, not in cat. The solution - using a template for the prefixes - is idiotic, but works. 2004-06-29 Kilian A. Foth * grammar/deutsch/makros.m4 (1.13): Consolidate special treatment of `statt'/`außer'. 2004-06-28 Olga Gertchikova * grammar/deutsch/Nomen.txt (1.105): the entry for "Nummer" changed 2004-06-28 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.143): rules and examples are improved 2004-06-28 Kilian A. Foth * utils/eval-tagger.pl (1.12): Order tag error summary alphabetically as well as by number for consistency. 2004-06-28 Kilian A. Foth * utils/tokenize.pl (1.34): Correctly tokenize `"sex and drugs"-Generation' into one word. 2004-06-28 Kilian A. Foth * libcdg/input.c (1.298): Increase minial length of words that may be used as unmarked suffixes from 3 to 4. Suffixes with 3 letters give way too many false positives, particularly things like `Ion', `Gen', `Ren' etc. Feel free to make the threshold user-settable if you don't like that. 2004-06-28 Kilian A. Foth * xcdg/parse.tcl (1.74): Fix getScore{} so that it actually does what the comment says. 2004-06-25 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.142), grammar/deutsch/Hierarchien.cdg (1.43), grammar/deutsch/Lexikon.cdg (1.121), grammar/deutsch/Nomen.txt (1.104), grammar/deutsch/Verben.txt (1.118): - random new lexicon items - obey leftpenalty only for non-verb regents; in subclauses you have no choice but to subordinate on the left. Hence: kein Bier _mehr_ *_mehr_ kein Bier But: wir gewinnen nicht _mehr_ seit wir nicht _mehr_ gewinnen - distinguish between sentence scope in the broader and the narrower sense - `2.' can be ADJD as well as ADJA - demand comma for subordinated S just like jor OBJC 2004-06-24 Kilian A. Foth * libcdg/compile.c (1.82), libcdg/predicates.c (1.68): Fix minor error in predHas(): -1 should be `no limit' both for the `min' and the `max' argument. This has NOT been corrected in compile.c, only noted! Please fix! 2004-06-23 Kilian A. Foth * xcdg/databrowser.tcl (1.10): Add keypress sensitivity to the grammar element browsers: just type `Ne[TAB]', and the program will select the first of your many `Nebensatz' constraints, and scroll you there so you can see what to type next. 2004-06-22 Kilian A. Foth * libcdg/lexemgraph.c (1.137): Do not attempt to look up `title-case' variants for words with only one letter. 2004-06-22 Kilian A. Foth * xcdg/allconstraints.tcl (1.32): Add `Weight' button, so you can change the weight of constraints that the shell refuses to have typed into it. 2004-06-21 Kilian A. Foth * configure.in (1.27): Remove stray reference to grammar/heise/Makefile. 2004-06-18 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.141), grammar/deutsch/Hierarchien.cdg (1.42), grammar/deutsch/Lexikon.cdg (1.120), grammar/deutsch/Nomen.txt (1.103), grammar/deutsch/Verben.txt (1.117), grammar/deutsch/doc.tex (1.50), grammar/negra/known_errors (1.45): - random new lexicon items - tidy up Projektivität1 and Projektivität2; the other four are still ugly monstrosities - standardize the analysis of `mit drei bis vier Mitarbeitern' - forbid reflexive pronouns without corresponding subjects - normalize `kurz vor dem Ende' etc.: PP now always modifies the adjective - forbid appositions under attributes - cardinal nouns are ATTR, not DET: *die zufriedene Kunden die zufriedenen Kunden Millionen zufriedene Kunden *Millionen zufriedenen Kunden - try to prevent common misparsing of subclauses as coordinations: [VVFIN]\ / \ KON / \ / [VVFIN] [ADV] \ \ OBJA \ [NN] should usually be [VVFIN] / \ NEB / \ OBJA / \ [VVFIN] [NN] / / / [ADV] - ensure that `aber' etc. are only ADV if they really need to, i.e. they actually are inverted - distinguish personal and reflexive pronouns everywhere - disprefer PTKVZ that look like APZR - restore broken sein-Subjekt-fehlt - fix comment at VOK-Definition 2004-06-18 Kilian A. Foth * utils/ccc-Makefile (1.9): Do not tokenize automatically. This is much too likely to change the delimitation of sentences, which should be done only on sober reflection. Use `make tokenize' manually now if needed. 2004-06-18 Kilian A. Foth * utils/ttc-wrapper.pl (1.1): New home for a script which is really corpus-independent. 2004-06-18 Kilian A. Foth * grammar/Makefile.am (1.12): Move heise corpus to /data/corpus. 2004-06-18 Kilian A. Foth * utils/search-annotations.pl (1.43): Default to searching `find|grep cda$` if no file given. 2004-06-18 Kilian A. Foth * utils/posdebug.pl (1.6): Ignore quotation marks like deutsch-tagger.pl does. 2004-06-18 Kilian A. Foth * utils/eval-tagger.pl (1.11): Sort both the corpus and the alternatives for one word consistently. 2004-06-18 Kilian A. Foth * utils/ccc-renumber (1.11): Document command-line options. 2004-06-18 Kilian A. Foth * utils/ccc-check (1.14): Use part-of-speech tagging with option -p (usually off to save much time). 2004-06-18 Kilian A. Foth * xcdg/allhierarchies.tcl (1.30): Change placing of items in trees so that you can actually read all of them. 2004-06-18 Kilian A. Foth * xcdg/main.tcl (1.59), xcdg/balloon.tcl (1.6), xcdg/parsetree.tcl (1.111): Move word tooltip to a place where it does not obscure the word. 2004-06-17 Othello Maurer * libcdg/predicates.c (1.67): - more debug output for 'predHasImpl()' 2004-06-17 Othello Maurer * libcdg/compile.c (1.81): - bug fixing. the compiler module may be working properly now 2004-06-17 Othello Maurer * libcdg/predicates.c (1.66), libcdg/predicates.h (1.22): - removed the old 'predHas()' and 'predBetween()' predicates. the new versions of predicates together with their 'impl()' functions seem to give the same results as the old ones 2004-06-17 Othello Maurer * libcdg/transform.c (1.37): - sort conflicts on output 2004-06-16 Olga Gertchikova * grammar/deutsch/Nomen.txt (1.102): declination of 'Mund' improved 2004-06-16 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.140): rules and examples impoved 2004-06-16 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.139): examples improved 2004-06-16 Kilian A. Foth * utils/ccc-find-conflicts (1.3): Fix octal number suppression: Tcl chokes on `009', so we translate it to `9', but we also changed `000' to `'. 2004-06-16 Kilian A. Foth * grammar/negra/stats.pl (1.9): Adapt to new sentence names. 2004-06-16 Kilian A. Foth * grammar/negra/compare.sh (1.15): Update tagger invocation. 2004-06-16 Kilian A. Foth * xcdg/parsetree.tcl (1.110): With left justification, manual filling with spaces is no longer necessary. 2004-06-16 Kilian A. Foth * utils/ccc-check (1.13): Option -t sets the threshold of ignored conflicts (in case you want all conflicts instead of just the serious ones). 2004-06-16 Kilian A. Foth * utils/cdg (1.20): Do not pass `quit' commands through to xcdg, where they make no sense. 2004-06-15 Kilian A. Foth * libcdg/functions.c (1.46), libcdg/functions.h (1.21), doc/doc.tex (1.84): New function `parent' returns the regent of an arbitrary word. Not actually used, since the rules I wanted to use it for turned out to be ineffective (see example in doc.tex), but a convenient function. 2004-06-15 Kilian A. Foth * utils/shift-reduce.pl (1.12): Option -f may specify a file of category equivalences used for tying parameters together. 2004-06-15 Kilian A. Foth * xcdg/main.tcl (1.58), xcdg/parse.tcl (1.73), xcdg/parsetree.tcl (1.109), libcdg/parse.c (1.183), libcdg/parse.h (1.60): Tooltips for words display the Value of the lexicon item, not just the description (which is usually incomplete). 2004-06-14 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.138): examples changed 2004-06-14 Lidia Khmylko * grammar/deutsch/Lexikon.cdg (1.119): the adverb 'genau' added; the definition of 'dort' improved 2004-06-14 Lidia Khmylko * grammar/deutsch/Adjektive.txt (1.82): lexeme added 2004-06-14 Kilian A. Foth * libcdg/frobbing.c (1.229): Add missing `determineAmbiguity()' near the entry point (could lead frobbing to give up too soon in additional runs). 2004-06-11 Kilian A. Foth * xcdg/commands.tcl (1.116), xcdg/menu.tcl (1.33), xcdg/shell.tcl (1.58): Optionally suppress the `Really exit?' dialog. 2004-06-10 Kilian A. Foth * README (1.5): Slightly update beginners' instructions. 2004-06-10 Kilian A. Foth * utils/scripts/newnet (1.1), utils/scripts/showparse (1.1), utils/scripts/trinamic (1.1): Some more useful examples of cdg scripts: newnet builds a net showparse immediately pops up a tree trinamic calls dynamic frobbing, and then another global frobbing (as used for COLING paper) 2004-06-10 Kilian A. Foth * libcdg/input.c (1.297): Fix automatic anaysis of marked compounds: only assume a compound when both the prefix and the suffix are nonempty. (`Code-freie' is ADJA, but `-freie' is now unambiguously TRUNC, never ADJA. Fight ambiguity!) 2004-06-10 Michael Daum * grammar/deutsch/Makefile.am (1.23): - fixed Names.cdg target to set the CDGHOME as all other perlscripts do 2004-06-09 Kilian A. Foth * libcdg/netsearch.c (1.116): Let netsearch post mean results like frobbing does. (Of course, there are rarely any above 0.0, so they don't pop up automatically anyway.) 2004-06-09 Kilian A. Foth * grammar/negra/known_errors (1.44), grammar/deutsch/Adjektive.txt (1.81), grammar/deutsch/Grammatik.cdg (1.137), grammar/deutsch/Namen.txt (1.64), grammar/deutsch/Nomen.txt (1.101): - random new lexicon items - fix type in find_relative - try harder to detect parallel indirect and direct left-subordination (which violates the Vorfeld structure, but cannot be seen by constraint `Vorfeld') - do not punish rhetorical `oder?' - reshuffle GMOD case constraints (but still a horrible mess) - correct table name in 'adverbiale PP nach Nomen' - allow even unusual numbers as ZEIT - forbid definite article below GRAD: `viele Meter große Fahrzeuge' `die Meter großen Fahrzeuge' *`die Meter große Fahrzeuge' 2004-06-09 Kilian A. Foth * utils/ccc-renumber (1.10): Move newly-written annotations to their proper subdirs via system('mv') rather than opendir(). 2004-06-09 Kilian A. Foth * utils/ccc-parse (1.10): Use subdirectories iff they are present, not when `-n' is not given (i.e. always do the Right Thing rather than forcing the user to remember a switch). 2004-06-09 Kilian A. Foth * utils/ccc-show-conflicts (1.7): Try to remove a common xhost setting that would spook xcdgclient.tcl. 2004-06-09 Kilian A. Foth * libcdg/frobbing.c (1.228), libcdg/netsearch.c (1.115), libcdg/command.c (1.306), libcdg/netsearch.h (1.33), doc/doc.tex (1.83): Overhaul option `cheatmode': - rename to `autocompare', since it has nothing to do with cheating. - let frobbing react to it as well as netsearch. - fix crippling bug where net->lg->isDeletedNode was accessed whilee that field still pointed to a stale bute vector. 2004-06-09 Kilian A. Foth * libcdg/compile.c (1.80): - do not declare `static ValueStruct' if no variable names follow - add `-I$CDGHOME' (how on Earth did you find config.h before?) - comTranslateBetween() is still wrong: it mis-translates even {X:SYN} : foo : exist : 0.01 : between(1, 2, ','); 2004-06-09 Kilian A. Foth * libcdg/predicates.c (1.65): Fix mistyped predBetweenImpl(). 2004-06-09 Kilian A. Foth * libcdg/predicates.c (1.64): between() now takes an inclusive range of word numbers so it works exactly the same as has(). 2004-06-09 Kilian A. Foth * libcdg/input.c (1.296), libcdg/input.h (1.136): Export freeAnnoEntry (used by ccc-renumber to save memory). 2004-06-08 Othello Maurer * libcdg/compile.c (1.79): - fixed wrong treatment of unsymmetric connexions - added 'parens' function - fixed wrong key construction in function _lookup the compile module seems to be working now. 'testing 4' is running without any errors... 2004-06-08 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.136): rules improved 2004-06-08 Lidia Khmylko * grammar/deutsch/Verben.txt (1.116): a verb added 2004-06-08 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.135): Examples improved, and also the rules for "Nebensatzstellung", "Infinitiv-Letztstellung", "PWAV-Kategorie" 2004-06-07 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.133): spelling of examples corrected 2004-06-07 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.134): [no log message] 2004-06-07 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.80), grammar/deutsch/Grammatik.cdg (1.132), grammar/deutsch/Lexikon.cdg (1.118), grammar/deutsch/Namen.txt (1.63), grammar/deutsch/Nomen.txt (1.100), grammar/deutsch/Verben.txt (1.115), grammar/deutsch/doc.tex (1.49), grammar/deutsch/makros.m4 (1.12), grammar/negra/known_errors (1.43): - random new lexicon items - try to disambiguate space and time adjuncts: `gestern auf der COLING' - `mehrere' is called a pronoun by STTS, but it does not govern adjectives like a pronoun; so make an exception for it. - fix `NN mit PP beigeordnet' - fix the various variants of `PRED-ADV-Reihenfolge' - unify handling of `initial' components (those that violate ordering constraints because they contain relative pronouns etc.) to a macro 2004-06-07 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.41): Create secondary passive participals: `auszugebenden' 2004-06-07 Kilian A. Foth * utils/tokenize.pl (1.33): Improve end-of-sentence detection. 2004-06-07 Kilian A. Foth * utils/ccc-renumber (1.9): Deal correctly with corpora split into subdirs. 2004-06-07 Kilian A. Foth * utils/ccc-check (1.12): Load all annotations in bulk; this is twice as fast as constantly exercising the autoloader. 2004-06-07 Kilian A. Foth * bindings/CDG/Utils.pm (1.8): - write_anno(): optionally write to a different name - format_string(): protect CDG keywords 2004-06-03 Michael Daum * grammar/negra/negraplugin.pl (1.73): - adapted to recent callback interface changes 2004-06-03 Michael Daum * utils/depsy.pl (1.7): - the phrase- and depConvTables now map a _list_ of perl functions to a tag; all functions are called in the given order 2004-06-02 Lidia Khmylko * grammar/deutsch/Namen.txt (1.62): a surname added 2004-06-02 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.131): several examples improved, the rules "Anrede ohne Komma 2/3/4" fixed 2004-06-02 Kilian A. Foth * utils/ccc-check (1.11): Ensure that the tagger is off. 2004-06-02 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.79), grammar/deutsch/Grammatik.cdg (1.130), grammar/deutsch/Lexikon.cdg (1.117), grammar/deutsch/Nomen.txt (1.99), grammar/deutsch/Verben.txt (1.114), grammar/deutsch/make-names.pl (1.15), grammar/negra/known_errors (1.42): - random new lexicon items - demand subject for `sein' except with impersonal adjectives: `Mir ist kalt.' *`Mir ist froh.' - force parenthesized nouns to be APP: `Das Forschungsministerium (BMFW) gab den Bericht heute bekannt.' *`Das Forschungsministerium gab (den Bericht) heute bekannt.' - treat , as a barrier for APP like for KON - fix and remove redundant and wrong contraints - allow coordination of GMOD and PP+`von' - disprefer S before & under adjective: `Ich war sicher, mich nicht verhört zu haben.' *`Ich war, mich nicht verhört zu haben, sicher.' - forbid multiple PART subordinations - forbid split infinitives - forbid modification of `man', `wer/PREL' etc. - correct wrong constraint about order of imperatives - normalize `ein Jahr nach der Katastrophe': the noun is now always the head word. 2004-06-02 Kilian A. Foth * utils/tokenize.pl (1.32): - optional parentheses debugging with -d - suppress the sentence prefixes under -s (useful to get more accurate diffs when changing the program changes the sentence delimitation) - paren counting can now deal with the most common emoticons - do not break at : within parentheses - handle abbreviations with more than one full stop: S.W.I.F.T. 2004-06-02 Kilian A. Foth * utils/ccc-show-conflicts (1.6): Use `cdg -x' instead of `xcdg', since not everyone has that link. 2004-06-02 Kilian A. Foth * utils/ccc-check (1.10): Handle corpora with `000' subdirs. 2004-06-02 Kilian A. Foth * utils/ccc-Makefile (1.8): Optionally, run a permission fixer after the main build. 2004-06-02 Kilian A. Foth * xcdg/parse.tcl (1.72): Change obsolete reference to levelPrio. 2004-06-02 Kilian A. Foth * libcdg/command.c (1.305), grammar/deutsch/base.m4 (1.19): Change tagger command: you MUST now say `tagger on' or `tagger off'. Previously, a solitary `tagger' in your .cdgrc and in deutsch.cdg would cancel each other. 2004-06-02 Kilian A. Foth * doc/doc.tex (1.82): Change tagger command: you MUST now say `tagger on' or `tagger off'. P reviously, a solitary `tagger' in your .cdgrc and in deutsch.cdg would cancel each other. 2004-06-02 Michael Daum * grammar/negra/negra.diff (1.2): updated our negra diffs 2004-06-02 Kilian A. Foth * grammar/deutsch/extra.m4 (1.13): Allow names with embedded upcase. 2004-06-02 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.40): Improve names of auto-generated valence constraints, and ensure that only one `OBJ not allowed' fails at a time. 2004-06-01 Olga Gertchikova * grammar/deutsch/Nomen.txt (1.98): a new entry 2004-06-01 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.129): examples changed 2004-06-01 Lidia Khmylko * grammar/deutsch/Nomen.txt (1.97): 'sto' - feature to the words mit -Wolle added 2004-06-01 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.128): examples changed 2004-06-01 Othello Maurer * libcdg/compile.c (1.78): - deactivated printing of 'type mismatch' errors in unequations 2004-06-01 Othello Maurer * libcdg/compile.c (1.77): [no log message] 2004-06-01 Othello Maurer * libcdg/compile.c (1.76): - corrected error in 'phraseqoutes' function - added handling for 2nd term of type 'RTString' in 'match' function - added special handling for chunk_start, chunk_end, chunk_type attributes 2004-06-01 Kilian A. Foth * libcdg/frobbing.c (1.227): manual(): move `t' keybinding from `show Time' to `set ignoreThreshold'. 2004-06-01 Kilian A. Foth * libcdg/tagger.h (1.16): Remove wrong comment. 2004-06-01 Kilian A. Foth * libcdg/command.c (1.304), libcdg/constraintnet.c (1.174), libcdg/input.c (1.295), libcdg/input.h (1.135): Abolish InputStruct->LevelPrio. Since Levels are kept in a list, we don't need another structure just to impose an ordering on them. 2004-06-01 Kilian A. Foth * xcdg/Cdgrc (1.29), xcdg/parsetree.tcl (1.108): Let parse trees actually obey the colour resources in Cdgrc. Edges highlighted when clicking conflicts are now highlighted in reddish-green, depending on the seriousness. Edges highlighted by `show cycles' etc. are highlighted in the -errorcolor (default red). Edges touched or highlighted by `showparse s1:4' are highlighted in the -highlightcolor (default white). You can get the old behaviour by setting *ParseTree.highlightColor to `red'. How do you set the background color of the parse tree canvas? 2004-05-28 Lidia Khmylko * grammar/deutsch/Lexikon.cdg (1.116): 'nur' has rightpenalty 2004-05-28 Othello Maurer * libcdg/compile.c (1.75): - added 'com->currentFormula' field for the formula currently beeing translated - added helper function 'comEscapeQuotes' for escaping Quotes in Strings which normally cause syntax errors when they are printed unescaped. Often this is the matter with 'com->currentConstraint->id' - added some more global static variables to the translated c file - implemented the 'has' predicate TODO: the 'match' function does not handle a third argument of type 'TTString' yet 2004-05-28 Othello Maurer * libcdg/predicates.c (1.63): - slightly changed order of elements in argument array for predHasImpl 2004-05-28 Kilian A. Foth * xcdg/allconstraints.tcl (1.31), libcdg/cdg.c (1.85), libcdg/cdg.h (1.63), libcdg/input.c (1.294), doc/doc.tex (1.81): Fix broken `Show' button for constraints, and make "a encoding user-settable. 2004-05-28 Kilian A. Foth * xcdg/visparses.tcl (1.47): Fix broken `refresh' menu item. 2004-05-26 Lidia Khmylko * grammar/deutsch/Nomen.txt (1.96): [no log message] 2004-05-26 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.126): example changes 2004-05-26 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.127): examle sentences changed 2004-05-26 Olga Gertchikova * grammar/deutsch/Lexikon.cdg (1.115): baseform for 'sämtlich' added 2004-05-26 Kilian A. Foth * doc/doc.tex (1.80), libcdg/functions.c (1.45), libcdg/functions.h (1.20): Function `parens' counts parentheses like `quotes' counts quotation marks. 2004-05-26 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.78), grammar/deutsch/Grammatik.cdg (1.125), grammar/deutsch/Lexikon.cdg (1.114), grammar/deutsch/Nomen.txt (1.95), grammar/deutsch/Verben.txt (1.113), grammar/deutsch/doc.tex (1.48), grammar/negra/known_errors (1.41): - random new lexicon items - allow `so Schröder' as a PAR constituent - disprefer relative clauses for short PP kernels - disprefer fragmentary subclauses even with KON - disprefer isolated ZEIT nouns 2004-05-25 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.124): Spelling and examples corrected 2004-05-25 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.123), grammar/deutsch/Lexikon.cdg (1.113), grammar/deutsch/Namen.txt (1.61), grammar/deutsch/Nomen.txt (1.94), grammar/deutsch/Verben.txt (1.112): rules corrected 2004-05-25 Kilian A. Foth * utils/ccc-check (1.9): Do not eval if the grammar has not changed since the last eval. 2004-05-25 Kilian A. Foth * utils/ccc-Makefile (1.7): Fix erroneous explanation of `quick' target. 2004-05-25 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.122): syntax fixed 2004-05-24 Kilian A. Foth * xcdg/allparses.tcl (1.60): Show the search method next to the comment in the parses table. (This is much more useful, since it is always set while `comment' is always empty.) 2004-05-24 Kilian A. Foth * libcdg/testing.c (1.41), libcdg/testing.h (1.6): tst3() detects instances of constraints that fail when subtrees of a complete tree are treated as fragments. 2004-05-24 Kilian A. Foth * grammar/negra/known_errors (1.40), grammar/deutsch/Grammatik.cdg (1.121), grammar/deutsch/Hierarchien.cdg (1.41), grammar/deutsch/Lexikon.cdg (1.112), grammar/deutsch/Nomen.txt (1.93), grammar/deutsch/Verben.txt (1.111): - random new lexicon items - fix several example sentences and constraints so that the examples actually exercise the constraints they explain - disallow ZEIT nouns to take non-ZEIT coordinations - disprefer adjectives as PN when used with nouns: für die Zukunft planen ein Plan für die Zukunft für gut halten *eine Haltung für gut - forbid modification of PTKZU - forbid modification in front of a free (conjunction-less) subclause - disambiguate `er hat es verstanden, zu gefallen': EXPL must modify the full verb - demand comma not only between the ends of a KON edge, but between children of the upper word and the lower word 2004-05-24 Kilian A. Foth * grammar/deutsch/doc.tex (1.47): Fix typing errors. 2004-05-24 Kilian A. Foth * utils/ccc-Makefile (1.6): Allow corpus-specific configuration via local file `OPTIONS'. 2004-05-24 Kilian A. Foth * utils/ccc-show-conflicts (1.5): Do not assume that xcdgclent.tcl is in $PATH. 2004-05-21 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.120): Fix over-complicated and wrong has() usage in two formulas. 2004-05-21 Kilian A. Foth * test/Makefile.am (1.23): It turns out that libcdg/libcdg.la is needed to link test executables, but specifying it in LD_ADD not only adds it (correctly) to the libtool invocation, but also creates a wrong dependency in the resulting Makefile. Therefore it must be present. 2004-05-21 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.119), grammar/deutsch/Nomen.txt (1.92), grammar/deutsch/Verben.txt (1.110), grammar/deutsch/doc.tex (1.46): - random new lexicon items - disprefer adverbial participles that look like auxiliary groups - allow nominalization only between phrase-quoted and non-phrasequoted words Rationale: it is not enough to check that the nominalizes word is in phrase quotes. This would mean that in the sentence "Ich weiß nicht, was ich noch machen soll" sagte sie. all of the quoted words could be e.g. appositions to each other, when actually we only want to allow the subordination *across* the quotation mark. - split `SVO im Nebensatz' into two rules, one triggering on relative pronouns, one on conjunctions - normalize long ZEIT attachments to use APP internally (was already done but incompletely) 2004-05-21 Kilian A. Foth * grammar/deutsch/Lexikon.cdg (1.111): `manch' can be plural. 2004-05-21 Kilian A. Foth * grammar/deutsch/TODO (1.7): Delete answered question. 2004-05-21 Kilian A. Foth * grammar/deutsch/Hierarchien.cdg (1.40): `APP' is a scope-preserving relation (has apparently always been missing). 2004-05-21 Kilian A. Foth * libcdg/command.c (1.303), libcdg/constraintnet.c (1.173), doc/libcdg.tex (1.91): Move documentation from .tex to .c. 2004-05-21 Kilian A. Foth * libcdg/frobbing.c (1.226): manual(): `r' issues a renewnet, not just a `reset history' 2004-05-21 Kilian A. Foth * libcdg/functions.c (1.44): funcPhrasequotes(): print the constraint that causes the error 2004-05-21 Kilian A. Foth * libcdg/input.c (1.293): - use hashForEachFreeValue() instead of local macros - fix invalid deletion of inputCompoundCats 2004-05-21 Kilian A. Foth * libcdg/predicates.c (1.62): Finetune the behaviour of has() with range arguments to the semantics that is actually needed: - both figures are inclusive: has(... ,1,3) checks three timepoints, not two - -1 means lg->min or lg->max - max < min is allowed: has(..., 3, 1) always fails 2004-05-21 Kilian A. Foth * libcontainer/blah.h (1.10), libcontainer/hashtable.c (1.5): hashForeachFreeValue() deletes a hashtable and all of its values, but not the keys. This is what we usually need, since the keys are Strings owned by the String store. 2004-05-21 Kilian A. Foth * test/Makefile.am (1.22): Remove dependency to libcdg/libcdg.la. (Although this file is reconstructed all the time it never actully changes, and it causes vectortest.c etc. etc. to be recompiled whenever frobbing.c changes, which is nopnsensical.) 2004-05-21 Kilian A. Foth * utils/ccc-Makefile (1.5): Introduce dependency to the base files constituting the deutsch grammar. 2004-05-21 Kilian A. Foth * utils/ccc-check (1.8): Introduce progress indicator. 2004-05-19 Kilian A. Foth * libcdg/tagger.c (1.58): Write forgotten taggerFinalize() (empty, only used under #define SERIOUS_FINALIZATION). 2004-05-18 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.118): syntax fixed; bad examples were marked with a sentence "ein schlechtes Beispiel;" 2004-05-18 Kilian A. Foth * utils/ccc-parse (1.9): Option `n' says `no recursion, store everything in . instead of 000/ etc.' 2004-05-17 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.117): syntax fiwed 2004-05-14 Kilian A. Foth * libcdg/constraintnet.c (1.172), libcdg/parse.c (1.182), libcdg/parse.h (1.59): Clarify comments. 2004-05-14 Kilian A. Foth * libcontainer/blah.h (1.9), libcontainer/list.c (1.22): listDeepClone() clones a list AND its elements. 2004-05-14 Kilian A. Foth * grammar/deutsch/Nomen.txt (1.91), grammar/deutsch/Verben.txt (1.109), grammar/deutsch/extra.m4 (1.12): - random new lexicon items Well, not *totally* random this time. These are actually all previously unknown verbs that appear in the entire heiseticker corpus. 2004-05-14 Kilian A. Foth * grammar/deutsch/doc.tex (1.45): Clarify rules for dealing with ill-firmed input. 2004-05-14 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.116): - clarifications top constraints found by exercising them all systematically - demand a coordination for all TRUNC items - treat `statt' like a conjunction 2004-05-13 Othello Maurer * libcdg/predicates.c (1.61), libcdg/predicates.h (1.21): - split the 'has' predicate for making the implementation accessible for the compile module. -> the argument list for the predHasImpl() function would be huge, thus a pointer to an array holding all the parameters needed by the mplementation is passed. !! define the SPLITPREDICATES makro for using the predHasImpl function 2004-05-12 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.115), grammar/deutsch/Verben.txt (1.108): syntax fixed, valences added, case gen for some names added 2004-05-12 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.114): - distinguish `worden' from `geworden' 2004-05-12 Kilian A. Foth * grammar/deutsch/Hierarchien.cdg (1.39): `SUBJC' is NOT an intra-scope label, in fact I introduced SUBJC explicitly to distinguish between scope-preserving and not preserving subjects. How did this sneak back under `Skopus'?? 2004-05-12 Kilian A. Foth * grammar/deutsch/Lexikon.cdg (1.110), grammar/deutsch/Nomen.txt (1.90), grammar/deutsch/Verben.txt (1.107): Random new lexicon items. 2004-05-12 Kilian A. Foth * utils/ccc-Makefile (1.4): Distinguish `quick' and `$(BASE).cdg' preconditions. 2004-05-12 Kilian A. Foth * xcdg/allparses.tcl (1.59): Unconditionally show parses created during manual frobbing. If the user explicitly typed `^', we can safely assume that they actually wanted to see that tree even if it is not scored >0. 2004-05-12 Kilian A. Foth * grammar/negra/negraplugin.pl (1.72): Assign cat:PPER to pronouns split off clitic verbs ("gibt's"). 2004-05-12 Kilian A. Foth * libcdg/frobbing.c (1.225): manual frobbing: new `w' subcommand solves a subproblem like `method=dynamic' does. 2004-05-12 Kilian A. Foth * libcdg/transform.c (1.36): Fix optional debug code. 2004-05-12 Kilian A. Foth * grammar/deutsch/make-nouns.pl (1.24): Fix declination error for type m7. 2004-05-11 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.113): syntax fixed 2004-05-11 Dietmar Dreyer * yada/YadaRanking.tcl (1.17), yada/YadaDocument.tcl (1.11), yada/YadaPlainDocument.tcl (1.11), yada/YadaProc.tcl (1.20), yada/YadaRunner.tcl (1.26), yada/YadaExperiment.tcl (1.9), yada/YadaMain.tcl (1.17), yada/YadaRunnableDocument.tcl (1.17): Extended YadaRanking with persistancy functionality. YadaRunner (and base classes) modified to support delegation for loading data relevant to the ranking document. 2004-05-11 Michael Daum * libcdg/compile.c (1.74): - fixed comTranslateUnder to catch every compiletime error 2004-05-11 Othello Maurer * libcdg/compile.c (1.73): - added 'is' and 'connected' predicates - the module is compiling without any errors but there was yet no testing of the compiled constraint's runtime behavior. 2004-05-11 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.112): syntax fixed 2004-05-10 Kilian A. Foth * libcdg/compile.c (1.72): Fix compiler warnings. 2004-05-10 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.111): syntax fixed 2004-05-10 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.110): - reintroduce shallow preference for [CARD NN] as CARD-->NN - guard against nonexisting features in DET-SUBJ agreement rule - regulate comparison via KOKOM by categories binarily - restrict adverbial behaviour of participles to the ADJD reading - fix critical errors in `Wahrnehmungsverb ohne OBJA' and `SVO im Nebensatz' -- please apply this patch at once! 2004-05-10 Michael Daum * bindings/cdgtypemaps.i (1.9), bindings/swig_utils.h (1.5): - prefix for newer swigs 2004-05-10 Michael Daum * cdgp/Makefile.am (1.22), test/Makefile.am (1.21): - added libblah again to fix linking on debian/unstable 2004-05-07 Kilian A. Foth * libcdg/tagger.c (1.57): Believe it or not, the heiseticker corpus actually contains instances of ASCII 01 (in reporting the actual filename of a particular virus attachment). That means we can't use 1 for space escaping for the tagger -- we use 255 now. There is, of course, no guarantee that some obscure input might not contain ASCII 255 to represent for something, but I'm willing to wait until that happens to fix it properly. 2004-05-07 Kilian A. Foth * libcdg/lexemgraph.c (1.136), libcdg/lexemgraph.h (1.67), libcdg/tagger.c (1.56): lgQueryCat(): Do not try to reconstruct the list of possible categories for a word by inputLexiconGet(); simply read it off the already existing leaves opf the lexeme graph. (Through interaction of spurious uppercase and automatic compounding, the reconstruction was actually incorrect every 100,000 runs or so.) 2004-05-07 Kilian A. Foth * libcdg/frobbing.c (1.224): Export initialization of module-global variables to one function that is called after each entry into the module. 2004-05-06 Kilian A. Foth * utils/ccc-parse (1.8): Option -l parses only the specified list of sentence numbers. 2004-05-05 Othello Maurer * libcdg/compile.c (1.71): - implemented the following predicates and functions: Phrasequotes, Lookup, Quotes, Pts, Between - the implementation of the 'lookup' function was slightly more complex: -> there is a com->maxLookupStrings field for the max number of strings used for key construction in the lookup functions of any constraint. -> this number is recursively retrieved by using the new functions comMaxLookupStrings, comMaxLookupStringsInFormula, comMaxLookupStringsInTerm 2004-05-05 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.109), grammar/deutsch/Lexikon.cdg (1.109), grammar/deutsch/Verben.txt (1.106): - random new lexicon items - grammar buglet fixes - discourage adverbial participles that sound like auxiliary phrases 2004-05-05 Kilian A. Foth * utils/ccc-parse (1.7): Implement distributed annotation storage. Also, switch to perl from bash (lots of numeric calculations to be done). 2004-05-05 Kilian A. Foth * utils/ccc-rename (1.2): Catch files that exactly match the spec. 2004-05-05 Kilian A. Foth * libcdg/frobbing.c (1.223): Add more optional DEBUG info to combined(). 2004-05-05 Kilian A. Foth * libcdg/predicates.c (1.60): Fix subformula caching error: has() with more than four arguments cannot be cached even while the tree remains unchanged. 2004-05-05 Kilian A. Foth * libcdg/tagger.c (1.55): When failing to write tagger input file, admit to its name. 2004-05-04 Kilian A. Foth * xcdg/xcdg.tcl (1.91): Shorten URL in the banner message. 2004-05-04 Kilian A. Foth * grammar/deutsch/make-verbs.pl (1.39): Fix suppression of duplicate forms (can happen when verbs are semiregular, and the irregular paradigm has forms in common with the regular one). 2004-05-03 Othello Maurer * libcdg/predicates.c (1.59), libcdg/predicates.h (1.20): - split the 'between' predicate for using the implementation in the compile module 2004-05-03 Kilian A. Foth * grammar/deutsch/Nomen.txt (1.89): Fix declaration of `Atom'. 2004-04-30 Kilian A. Foth * doc/doc.tex (1.79): Document extension of findAnnotation(). 2004-04-30 Kilian A. Foth * libcdg/input.c (1.292): Extend annotation autoloading; when searching for foobar-s543876, not only the annodir is tried but also its `054/' subdirectory (because 543876 is in the 54th batch of 10,000). 2004-04-27 Dietmar Dreyer * yada/YadaRunner.tcl (1.25): Fixed bug occurring when loading runner document (trying to load file with invalid file name) 2004-04-26 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.108): syntax fixed 2004-04-26 Kilian A. Foth * doc/doc.tex (1.78), grammar/deutsch/Grammatik.cdg (1.107), grammar/deutsch/TODO (1.6), grammar/deutsch/doc.tex (1.44), libcdg/functions.c (1.43), libcdg/input.c (1.291), libcdg/predicates.c (1.58): Allow has() to take a range of numbers as the last parameters, which limit the range in which to search. Disclosure: I am contemplating yet another verb phrase reconstruction that will make it necessary to do just this to formulate the Vorfeld constraint. 2004-04-23 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.106): Add ancillary tagging constraints to prevent certain edges from being built int he first place, rather than building and deleting them again. 2004-04-23 Kilian A. Foth * libcdg/input.c (1.290): Restore umlauts lost in editor malfunction. Don't ask. 2004-04-22 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.105), grammar/deutsch/Nomen.txt (1.88), grammar/deutsch/TODO (1.5), grammar/deutsch/Verben.txt (1.105), grammar/deutsch/doc.tex (1.43): Random new lexicon items. 2004-04-22 Kilian A. Foth * doc/doc.tex (1.77), grammar/deutsch/Lexikon.cdg (1.108), grammar/deutsch/base.m4 (1.18), grammar/negra/known_errors (1.39), libcdg/input.c (1.289), libcdg/input.h (1.134): Implement selective use of lexicon templates. Rationale: So far the rule about lexicon templates was that they are only used if normal lexicon lookup fails entirely. Say that you have many nouns (NN), names (NE) and foreign-language words (FM) in your lexicon, but can't be sure that you have them all (in fact, you can be pretty sure that you don't). So you write lexicon templates for unknown nouns and names (basically matching all capitalized words) and for unknown foreign-language material (matching everything). Now you query the lexicon for `das', and it gives you items of type ART, PDS, and PRELS. Because the lookup succeeded, templates are not used. This is usually what you want - you are protected against insufficient coverage, but do not pay extra ambiguity if you do have coverage. The problem arises if you have coverage but of a wrong type. We have the German Adjective `golden'/ADJD, which means that the FM version is not returned even though the template would match. But what about the sentence `Der Staat hat sich einen golden share vorbehalten'? The foreign-language phrase cannot be analysed as an NP because `golden' is an ADJD, not an ADJA in German. The correct analysis is to treat it as two FM and chain them via APP; but `golden'/FM is blocked by the presence of the inappropriate `golden'/ADJA. The solution so far was to add `golden'/FM explicitly in the lexicon; but you can see that it is impossible to decide in advance which words might conceivably Therefore, the use of lexicon templates can now be regulated by setting the CDG variable `templates'. `never' and `always' do what they suggest. `ifneeded' selects the old behaviour. But the new and improved `bycategory' mode instructs the lexicon to use a template iff there is no normal entry of the same syntactic category. Thus `golden'/FM is allowed, and the sentence can be analysed correctly; but if you already have a noun reading, templates will not provide a redundant second one. The increased number of alternatives in the lexeme graphs could be a problem, since *everything* can now be FM. Luckily, you can usually rely on the tagger getting implausible categories right: it positively (score 1.0) asigns `golden' the category FM because the second FM `share' is a dead giveaway, and it positively rejects obviously wrong alternatives: `der' is never assumed to be FM. Therefore the ambiguity increase is checked at the cost of a slightly greater dependence on accurate POS tagging. A related change is that in the German grammar, all lexicon items will now be postfixed with their category, since they are all theoretically ambiguous w.r.t. category: `wird' becomes `wird_VAFIN' to distinguish it from the never-used `wird_FM'. This causes cosmetic but pervasive changes to all lists of known errors. Do not be afraid of the large diffs there. 2004-04-22 Kilian A. Foth * utils/ccc-parse (1.6): Option -r takes a range of sentence numbers to be processed. 2004-04-21 Kilian A. Foth * utils/ccc-check (1.7): Do not die unnecessarily when unlink() fails. 2004-04-20 Kilian A. Foth * utils/deutsch-tagger.pl (1.12): Detect `soweit'/KOUS misanalysed as ADV. 2004-04-20 Kilian A. Foth * utils/ccc-check (1.6): Remove wrong [shift] key event. 2004-04-20 Kilian A. Foth * grammar/deutsch/Lexikon.cdg (1.107), grammar/deutsch/Nomen.txt (1.87), grammar/deutsch/Verben.txt (1.104), grammar/deutsch/doc.tex (1.42): Random new lexicon items. 2004-04-20 Kilian A. Foth * grammar/deutsch/Grammatik.cdg (1.104): - demand number agreement for the determiners of subjects - demand case agreement for the determiners of PN edges under verbs - forbid the configuration X--OBJA-->Y--PN-->Z 2004-04-19 Lidia Khmylko * grammar/deutsch/Namen.txt (1.60), grammar/deutsch/Nomen.txt (1.86): lexics fixed 2004-04-19 Lidia Khmylko * grammar/deutsch/Lexikon.cdg (1.106): flexion of 'manch' fixed 2004-04-19 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.102): spelling fixed 2004-04-19 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.103): spelling fixed 2004-04-19 Kilian A. Foth * grammar/deutsch/make-names.pl (1.14): Use proper CDG parsing, not perl regexes, to determine the set of words for -c. 2004-04-19 Michael Daum * libcontainer/string.c (1.10): - removed debug switch 2004-04-19 Kilian A. Foth * libcontainer/string.c (1.9), libcdg/command.c (1.302): Only actually call strDecode() under XCDG. At the moment, iconv() does the right thing when called with wrong input encodings and leaves the string unchanged, but why tempt fate? 2004-04-19 Michael Daum * xcdg/shell.tcl (1.57): - fixed cdgGetString/shell::fgets 2004-04-19 Kilian A. Foth * utils/ccc-parse (1.5): Grep the actual indices of lattices in the input, not just their total number. 2004-04-19 Kilian A. Foth * libcdg/input.c (1.288): printIdentifier(): protect CDG keywords and backslashes so that writeAnnotation() always yields valid CDG input. 2004-04-16 Kilian A. Foth * libcdg/chart.c (1.6), libcdg/command.c (1.301), libcdg/eval.c (1.109), libcdg/frobbing.c (1.222), libcdg/functions.c (1.42), libcdg/increment.c (1.71), libcdg/input.c (1.287), libcdg/input.h (1.133), libcdg/lexemgraph.c (1.135), libcdg/parse.c (1.181), libcdg/transform.c (1.35), yada/YadaExperiment.tcl (1.8), yada/YadaExperiments.tcl (1.8), yada/YadaRunnableDocument.tcl (1.16): Correct spelling. 2004-04-16 Kilian A. Foth * grammar/deutsch/make-names.pl (1.13): Allow multiple genitives: `Lynch' neuer Film' vs. `Frankreichs Premier' 2004-04-16 Kilian A. Foth * grammar/deutsch/Adjektive.txt (1.77): Remove duplicate entries for words ending in `e'. 2004-04-13 Olga Gertchikova * grammar/deutsch/Verben.txt (1.103): the verb "vertrauen" valency fixed 2004-04-13 Olga Gertchikova * grammar/deutsch/Grammatik.cdg (1.101): spelling mistakes fixed 2004-04-13 Lidia Khmylko * grammar/deutsch/Namen.txt (1.59): lexics 2004-04-13 Lidia Khmylko * grammar/deutsch/Grammatik.cdg (1.100): spelling mistakes fixed 2004-04-13 Lidia Khmylko * grammar/deutsch/Adjektive.txt (1.76), grammar/deutsch/Namen.txt (1.58), grammar/deutsch/Nomen.txt (1.85), grammar/deutsch/Verben.txt (1.102): lexics 2004-04-13 Dietmar Dreyer * yada/YadaExperiment.tcl (1.7): Fixed bug concerning saving the active flag correctly 2004-04-13 Othello Maurer * libcdg/eval.c (1.108): - added argument 'context' to compiled constraint function calls 2004-04-13 Othello Maurer * libcdg/compile.c (1.70): - implemented the 'under' predicate - fixed some segfault errors: memory was freed which was actually owned by registered strings - changed names of generated constraint-functions: -> don't use constraint-id 's as function names anymore because they may contain characters (like spaces or quotes) which are not allowed to be part of function names. -> use constraint->no 's instead 2004-04-13 Othello Maurer * libcdg/predicates.c (1.57), libcdg/predicates.h (1.19): - moved the implementation of the predUnder() predicate to a seperate function - typechecking is done in the predUnder() function - the actual implementation moved to predUnderImpl() -> thus the implementation of the predicate can be used by compiled constraints - this functionality is deactivated right now but can be turned on by defining the SPLITPREDICATES macro 2004-04-13 Dietmar Dreyer * yada/YadaProc.tcl (1.19): Added Methods for busy mouse pointer 2004-04-08 Kilian A. Foth * utils/ccc-check (1.5): Split cdgp operation into runs of 1000 items each, in case you need to check a million sentences or something. 2004-04-08 Kilian A. Foth * utils/ccc-parse (1.4): Options `-e' or `-o' select only even/odd sentences for a run (useful for dual processor machines). 2004-04-06 Michael Daum * yada/YadaGrammars.tcl (1.25): - fixed choke if there's no nontcl grammar to load 2004-04-06 Michael Daum * grammar/deutsch/Makefile.am (1.22): - fixed %.dvi target (and all dependants) 2004-04-06 Michael Daum * utils/Makefile.am (1.22): - added depsy man page to install target