2004-01-12 20:26  kps

	* rigeail: Subtle portability bug -- on Darwin, "wc -c" leaves
	  trailing spaces after the number; different than GNU apparently.
	  Meant that aonchiall-ga.pl was being generated incorrectly,
	  leading to weird extra spaces in the output.

2004-01-12 20:24  kps

	* po/LINGUAS: Added "sk" and "mn"

2004-01-12 20:23  kps

	* cuidiu.html, foirm.html: Changes for Slovak/Mongolian
	  localizations.

2004-01-12 20:22  kps

	* po/fr.po: Quick turnaround on the 0.4 french po by Michel
	  Robataille.

2004-01-12 13:04  kps

	* stress, stress.err, stress.out: Added tests of localizations to
	  Mongolian and Slovak.

2004-01-11 21:41  kps

	* po/: mn.po, sk.po: Mongolian and Slovakian localizations; thanks
	  to Sanlig Badral <badral@users.sourceforge.net> and Andrej Kacian
	  <andrej@kacian.sk>, respectively.

2004-01-08 23:04  kps

	* stair.html: 15% improvement!

2004-01-08 23:02  kps

	* sios.html: Version bump.

2004-01-08 23:01  kps

	* stair.html: 0.3->0.4 changes added.

2004-01-08 22:33  kps

	* po/: de.po, fr.po, ga.po, gramadoir.pot, nl.po, ro.po: Time stamp
	  only.

2004-01-08 22:31  kps

	* po/: de.po, fr.po, ga.po, gramadoir.pot, nl.po, ro.po: Several
	  new msgids appear; some translator comments added.  New gettext
	  version adds "report msgid bugs" email address.

2004-01-08 22:28  kps

	* po/LINGUAS: add fr, nl, ro.

2004-01-08 22:28  kps

	* unigram-ga.txt: Some reordering; made use of p="xx" more
	  consistent for verb tags.

2004-01-08 22:27  kps

	* Makefile.in: triail.err no longer keeps line numbers.

2004-01-08 22:27  kps

	* triail.err: Additions, reshuffles, but main change is that I no
	  longer store the line numbers, to simplify reading future
	  "diff"'s!

2004-01-08 22:22  kps

	* stair.html: Version 0.4 benchmark data added.

2004-01-08 22:20  kps

	* sonrai.html: comhshuite "coming soon..."

2004-01-08 22:19  kps

	* rigeail: comhshuite changes.

2004-01-08 22:18  kps

	* rialacha-ga.in: Many changes/improvements: (1) System for "only
	  in set phrases" rules in place; more than 100 such added; (2)
	  Some tags added for efficiency (3) Rule req. genitive after
	  compound preps (4) lots of rules for using correct verbal
	  particles and correct induced initial mutations (5) improved
	  "cad" rules (6) small bug fixes (7) a couple of "CAIGHDEAN"
	  rules added explicitly (most others done in cuardach by lookup in
	  eile-ga.bs) -- e.g. frd

2004-01-08 22:11  kps

	* macra-ga.meta.sed: Many new macros added in conjuction with
	  increased complexity of rule sets.  Also changed many *'s to +'s
	  which helped efficiency a certain amount.

2004-01-08 22:10  kps

	* iompar.html: comhshuite-xx.in step added.

2004-01-08 22:10  kps

	* index.html: --aspell flag description added.

2004-01-08 22:10  kps

	* grweb: Web interface uses --aspell now.

2004-01-08 22:09  kps

	* gr.in: --aspell stuff added.	New messages for "set phrase" and
	  "genitive needed".

2004-01-08 22:07  kps

	* foirm.html: French localization added to web interface.

2004-01-08 22:06  kps

	* eisceacht-ga.in: Massive changes/additions/improvements.

2004-01-08 22:06  kps

	* debug.sed: Stuff added for (1) comhshuite and (2) because of
	  aspell changes

2004-01-08 22:05  kps

	* cuidiu.html: Moved contributor names to the top; added comhshuite
	  stuff.

2004-01-08 22:05  kps

	* configure.ac: Version bump to 0.4, comhshuite stuff added.

2004-01-08 22:04  kps

	* aonchiall-ga.in: Massive changes, improvements, bug fixes.

2004-01-08 22:03  kps

	* addlang: Comhshuite added.

2004-01-08 22:03  kps

	* Makefile.in: (1) stuff for building comhshuite-* (2) new "counts"
	  target for counting rule sets for web page

2004-01-08 21:01  kps

	* stress, stress.out: Added tests for the developer flags.

2004-01-08 16:21  kps

	* stress.err, stress.out: Output from latest version with aspell
	  checks in place.

2004-01-08 16:18  kps

	* stress: Two new checks: aspell stuff and sentence segmentation
	  check.

2004-01-06 20:25  kps

	* eagar.html, gramadoir.el: New version of emacs interface from
	  Martin Gregory, plus a nice README that I've marked up and
	  included in eagar.html.

2004-01-05 22:11  kps

	* po/fr.po: French localization, thanks to Michel Robitaille
	  <robitail@IRO.UMontreal.CA>.

2003-12-31 08:59  kps

	* stress.out: Improvements in tagging, renumbering of triail, etc.

2003-12-30 13:29  kps

	* Makefile.in: Another possibly unsafe temp file fixed (see
	  previous commit).

2003-12-30 13:26  kps

	* Makefile.in, rigeail: Subtle bug in build, discovered while
	  writing 0.4 Gentoo ebuild.  My multiprocessor machine
	  parallelizes the Makefile dependencies when possible; in
	  particular attempts to do some of the .in.pl files concurrently.
	  But these use(d) the same temp filename, causing problems.

2003-12-24 10:40  kps

	* rialacha-ga.in, triail.err: (1) renumbering of triail.err b/c of
	  move to rialacha-ga.in (2) New rules added: finally have "no /h/"
	  working on nouns, prefix-h after ordinal adjective, conjunction
	  "" lenites verbs (OD), first cut at "no lenition" rule on verbs.

2003-12-24 10:08  kps

	* unigram-ga.pre: Now all verb forms use "p="xx"" for generic case.

2003-12-24 10:06  kps

	* macra-ga.meta.sed: New macros.

2003-12-24 10:06  kps

	* eisceacht-ga.in: New rules; mostly "iff" work on prefix-h.

2003-12-24 10:05  kps

	* aonchiall-ga.in: Several new rules, some lumping together for
	  efficiency, and addition of "p="xx"" to generic present, future,
	  for consistency with past, cond., imperf.

2003-12-24 10:03  kps

	* TEANGACHA: Added Finnish (fi) locally.

2003-12-24 10:03  kps

	* Makefile.in: (1) new stair.html web page (2) Build "triail"
	  straight from comments in rialacha-ga.in

2003-12-23 23:47  kps

	* stair.html: Moved benchmark table up for easy comparison with
	  version table.

2003-12-23 23:46  kps

	* gr.in: 1. new output message 2. some tidying 3. Got rid of
	  "cipcheart" for use on web page; consistent with   gr --version
	  now.

2003-12-23 21:59  kps

	* rialacha-ga.in: Adjective lenition rules after prep+nb are
	  subsumed.

2003-12-23 21:02  kps

	* unigram-ga.pre: Another embarrassing bug which was causing the
	  "globbing" of past, conditional, etc. verb forms to fail before
	  unigram.

2003-12-23 11:39  kps

	* rialacha-ga.in: Added interleaved sample sentences from "triail"
	  to rialacha-ga.in.

2003-12-23 11:35  kps

	* triail: Moved sample sentences into rialacha-ga.in to speed up
	  development, testing.

2003-12-23 10:13  kps

	* rialacha-ga.in: First batch of LGG comments added.

2003-12-22 15:07  kps

	* rialacha-ga.in, triail, triail.err: New rules added: repetition
	  of definite article, "go+an"=> "go dt", "san+consonant", "trd
	  na"=>"tr na".  Added some <S>'s to "trd?" for efficiency since
	  this is the unigram tag anyway.

	  Corrected bug which reported "An aontaonn" as needing eclipsis
	  (it doesn't).

	  Unified and extended tag 2-grams involving <T>.

2003-12-18 12:31  kps

	* cuidiu.html, eagar.html, foirm.html, index.html, iompar.html,
	  samptemplate.html, sios.html, sonrai.html: Added links to
	  stair.html

2003-12-18 12:31  kps

	* stair.html: New release history/benchmark page.

2003-12-17 15:14  kps

	* addlang: New script for quickly adding a trivial language port.

2003-12-17 15:13  kps

	* foirm.html, gr.cgi, grweb: NLS added to web interface.

2003-12-17 15:12  kps

	* cuardach-en.c, eisceacht-en.in: Comment tweaks to help with
	  "addlang" script.

2003-12-17 15:11  kps

	* debug.sed: Added lines for "unigram" benchmark.

2003-12-12 19:00  kps

	* iompar.html: Huge changes based on all the stuff added for _true_
	  multilingual-ness, and experience in actually doing the ports to
	  "GA" (new char encoding) and "EN" (adding p.o.s. codes).

2003-12-12 18:59  kps

	* gr.in: 1. First cut at translator comments.  2. Adding correct
	  character encoding to xml stream now 3. Forgot the "toutf" with
	  --litriu flag.  4. One fewer "\n" in sed replacement!

2003-12-12 18:57  kps

	* cuardach-EN.c, unigram-EN.txt:
	  Only files (other than focail-EN.bs) distinct from the "en" in
	  the distribution.

2003-12-12 18:56  kps

	* eisceacht-en.in: Couple of cute examples in comments.

2003-12-12 18:51  kps

	* farc.c.in: Templates for new "tolower" calls in each language
	  file.

2003-12-12 18:51  kps

	* cuardach.c: Moved language specific "tolower" stuff out of here.

2003-12-12 18:50  kps

	* cuardach-en.c, cuardach-ga.c: Moved tolower stuff out of cuardach
	  into language files.

2003-12-12 18:49  kps

	* TEANGACHA, TEANGACHA.dist: ASCII datafile.  Also added "GA" to
	  local copy.

2003-12-12 18:48  kps

	* Makefile.in: - UTF8 version of "ga" added ("GA").  - farc.c stuff
	  for moving my_tolower out of cuardach.c

2003-12-12 13:02  kps

	* TEANGACHA: Changed to preferred "ISO-8859-1" as suggested by w3,
	  since this is now inserted into XML encoding specification.  Also
	  experimenting with EN as "ASCII" for testing purposes--- ispell
	  word list has no accented characters or hyphens in it!

2003-12-12 13:00  kps

	* TEANGACHA.dist: New file so that my local TEANGACHA can contain
	  experimental versions.  This will also be useful for building
	  specialized distributions as more languages become available.

2003-12-10 22:27  kps

	* cuardach.c: Silly bug found while playing with gettext.

2003-12-10 22:26  kps

	* Makefile.in, configure.ac: Changes corresponding to upgrade to
	  gettext-0.13.  See previous checkin of po/POTFILES.in, etc.

2003-12-10 22:25  kps

	* po/: Makefile.in.in, POTFILES.in: Changes corresponding to
	  upgrade to gettext-0.13.  Now can just use xgettext itself to
	  extract strings from "gr.in"	(though, in a bit of a hack, had to
	  (1) run configure so that @SH_GETTEXT@ gets substituted (2) get
	  rid of the "-s" flag on gettext which apparently  xgettext
	  doesn't recognize! (3) cp "gr" to "gr.sh" to fool xgettext into
	  recognizing it as "Shell"

2003-12-10 22:14  kps

	* macra-ga.meta.sed: Added possible h="y" attribute to all of the
	  ALLN* macros since w/o them, a lot of such words were ending up
	  as <V p="2" t="ord">, even after a preposition, because of the
	  unigrams (all of the h="y"'s are very low freq): "le hsid" etc.
	  These lead to false positives "unnec. prefix h"

2003-12-10 22:04  kps

	* gramadoir.el: Another improvement from Martin Gregory: highlights
	  the text in the original file as well as in the message buffer.

2003-12-09 23:45  kps

	* index.html: New command-line options.

2003-12-09 23:42  kps

	* gr.in: 1. Added Brill tagging algorithm, --brill option.  2.
	  Added NITEE message.	3. Got rid of remaining "\n" in sed
	  replacement strings, thanks to Martin Gregory for reminding me of
	  this problem with non-GNU versions (and even older GNU versions).

2003-12-09 23:23  kps

	* Makefile.in: Added TEANGACHA.dist stuff.  Added focail-EN.bs (in
	  addition to focail-en.bs).

2003-12-08 19:47  kps

	* rialacha-en.in: Forgot to add the "rare" rule to English dist.

2003-12-08 16:22  kps

	* triail, triail.err: Nine new sample sentences for latest batch of
	  rules added.	Also, last asterisk removed now that disambiguation
	  is "done".

2003-12-07 23:04  kps

	* stress.out: Tweaks reflecting recent changes: (1) new test for
	  segmentation of sentences (b/c of sed bug reported by Martin
	  Gregory) (2) Version bump (3) New copula tag appears once (4) New
	  command line options.

2003-12-07 15:00  kps

	* gramadoir.el:
	  New version of emacs interface from Martin Gregory.

	  Here's his description of the changes:

	  "1. it uses gettext to determine the correct version of the
	  "Currently checking" string and so works for any supported
	  locale;

	  2. it implements highlighting in the Gramadoir output buffer.
	  Here I have overcome the <b></b> problem you mentioned by
	  exploiting an Emacs feature called text properties, the
	  "invisible" property in particular. This means that I can still
	  find the tags (for the next stage of highlighting in the original
	  text - to be implemented next) but they are not visible.

	  3. it gracefully handles a call to next-message if there are no
	  more messages."

2003-12-07 14:54  kps

	* gr.in: Added new "developer" flags: --ilchiall, --minic,
	  --no-unigram, reorganized output of gr --help too.

2003-12-07 14:53  kps

	* Makefile.in: Might as well use the new minic flag.

2003-12-07 13:08  kps

	* rigeail: Tiny change with huge performance boost -- unigrams=>
	  can now assume no ambiguities when checking rialacha/eisceachta.

2003-12-06 15:21  kps

	* rialacha-ga.in:
	  1. "unnecessary h", "unnecessary t" rules 2. t- after "cn" 3.
	  nos, nb, nba, require comparative.  4. h after "p" 5. h after
	  "sna" 6. Bug in gender agreement rule gs+adj

	  Then some cosmetic changes because of new <[..]> syntax and some
	  added tags for efficiency.

2003-12-06 15:18  kps

	* iompar.html, sonrai.html: Unigram documentation.

2003-12-06 15:18  kps

	* stress: LC_ALL should suffice for setting output language,
	  affects nothing else.

2003-12-06 15:10  kps

	* macra-ga.meta.sed: Added some new macros, tweaked some old ones,
	  and changed some *'s to +'s for efficiency.

2003-12-06 15:09  kps

	* gin.meta.sed: (1) Moved macros to macra file.  (2) Added extended
	  syntax for regexp'd tags: <[ABC]>  <[^ABC]> etc.

2003-12-06 15:03  kps

	* eisceacht-ga.in: A few new exceptions added, in particular some
	  important ones which work in concert with strict "no-h" rules to
	  mark up unnecessary h's w/o false positives after "go, chomh",
	  etc.	 Planning lots more such rules for lenition, eclipsis, etc.

2003-12-06 14:53  kps

	* cuardach-ga.c: Added copula code/tag.

2003-12-06 14:52  kps

	* aonchiall-ga.in: More wholesale changes, taking us to first
	  version working in concert with unigram stuff to tag completely.

2003-12-06 14:48  kps

	* Makefile.in: Cleaned up dependencies for new unigram files.

2003-12-05 16:44  kps

	* unigram-ga.txt:
	  (1) added copula markup (2) bootstrap (reran it using "unigram"
	  as part of "gr") -- really this just boosts the frequent tags!

2003-12-05 11:30  kps

	* gr.in: (1) Deal with environment variables correctly -- tricky to
	  deal with desired output language/encoding (LC_CTYPE/LC_MESSAGES)
	  and also perform grammar checking correctly (2) Trap errors from
	  "iconv", avoids crash if not installed (3) toutf converts output
	  back to encoding of file (4) stuff from abairti about
	  ends-of-sentences after single letter words  moved into giorr-xx
	  as it is language specific (5) Unigram stuff added.

2003-12-05 11:25  kps

	* giorr-en, giorr-ga: Language-specific exceptions to "escape punc
	  after single character" -- no need for "a" in English since not
	  end of sentence, "I" appears as "why can't I?" etc.

2003-12-05 11:16  kps

	* cuidiu.html, iompar.html, sonrai.html: Web site changes: new
	  contributors, stuff about "rare" code 127, description of
	  "abairti" module and a mention of Brill stuff.

2003-12-05 11:14  kps

	* configure.ac: Unigram changes.

2003-12-05 11:04  kps

	* Makefile.in: Changes for unigrams.

2003-12-05 11:03  kps

	* unigram-en.pre, unigram-en.txt, unigram-ga.pre, unigram-ga.txt:
	  New "last resort" unigram disambiguation scheme.

2003-12-04 23:10  kps

	* triail.err: tr+a disambig => trna error now working in real
	  texts.

2003-12-04 13:46  kps

	* Makefile.in: forgot rigeail in tarball.

2003-12-04 12:04  kps

	* Makefile.in: (1) Stuff for language-dependent macro files
	  macra-xx.meta.sed.  (2) One strategic "LC_ALL=C" in the .in.pl
	  target which allows the build to proceed even if locale is utf8,
	  fixing bug submitted by Alan Horkan and David O'Callaghan.

2003-12-04 12:02  kps

	* stress.out: Trap iconv error at beginning and exit now! Also
	  added test of ga_IE.utf8.

2003-12-04 11:58  kps

	* stress.err: Trap "iconv" errors now.	Also added a test with
	  LC_ALL=ga_IE.utf8.

2003-12-04 09:12  kps

	* cabhair.c: 8-bit clean.

2003-12-03 15:08  kps

	* macra-en.meta.sed, macra-ga.meta.sed: New files, stripped out of
	  gin.meta.sed in an effort to modularize language-specific detail.

2003-12-03 15:07  kps

	* giorr-en: New file, copied from giorr-ga, stripped out Irish
	  stuff, and then added a few things based on running abbsurv on
	  some big texts from Project Gutenberg.

2003-12-03 14:34  kps

	* grweb: Only "LC_MESSAGES" was not sufficient for libc to
	  _display_ the gettext'd messages correctly; need LC_CTYPE too
	  (which I have in my environment anyway, web server user didn't!)

2003-12-03 14:30  kps

	* stress.out: Major overhaul.

2003-12-03 14:30  kps

	* stress.err: Added two languages.

2003-12-03 14:30  kps

	* stress: (1) smaller test set (overkill) (2) two new languages to
	  test (3) non-ascii example wasn't coming out since it ended in
	  "II" so that trailing punc was escaped!    Fixed now.

2003-12-03 13:56  kps

	* Makefile.in, gr.in: Changes from generalization of "giorr" to
	  "giorr-xx".

2003-12-03 13:31  kps

	* giorr-ga: Additions based on new language-independent "abbsurv"
	  script.

2003-12-03 08:47  kps

	* giorr-ga: Renamed from "giorr".

2003-12-03 08:46  kps

	* giorr: Renamed to giorr-ga.

2003-12-02 12:26  kps

	* aonchiall.meta.sed, rigeail: Tweak to the in->pl conversion so
	  that the "rigeail" backreference renumbering hack works
	  correctly.

2003-12-02 12:25  kps

	* sios.html: Gentoo moved to x86 stable!

2003-12-02 12:24  kps

	* gr.in: New UI messages.

2003-12-02 12:17  kps

	* giorr: Lots of additions from careful look at the corpus --wrote
	  script "abbsurv" which surveys tokens followed by ". ", filters
	  out correctly spelled words, sorts by frequency.

2003-12-02 12:15  kps

	* cuardach.c: Small bug (forgot about embedded newline in word) in
	  decapitalization.   Was treating ALLCAPS as an "unusual"
	  capitalization since the count of lowers was incremented at the
	  newline.

2003-12-02 12:13  kps

	* Makefile.in: Two small changes -- using ~ vs. ! as a dummy
	  character in making perl files so I can be excited in my *.in
	  comments !!

	  Also, important tweak to the construction of the focail-ga.bs
	  which forces the grammar code bytes to be sorted consistently so
	  that this can be exploited in writing disambiguation rules.

2003-12-02 12:11  kps

	* aonchiall-ga.in:
	  Massive additions to disambiguation module, mostly in the form of
	  Brill transformational rules deduced from the corpus.  This also
	  led to a non-trivial amount of hand-construction of rules for
	  dealing with the most frequent "genotypes" of ambiguities.

2003-11-25 22:46  kps

	* triail, triail.err: Added two new rules, p and sna PREFIXH.

2003-11-25 13:48  kps

	* giorr: Added most common abbreviations for the months.

2003-11-25 13:46  kps

	* rialacha-ga.in: (1) arna lenites (2) nos requires comparative
	  (3) generalized "huaire" rule to allow "ceithre uair" as error
	  e.g.	(4) MAIN CHANGE: addition of rules from analysis of 2-grams
	  of tags

2003-11-25 13:44  kps

	* triail.err: Found actual usages with "s uair" so generalized
	  rule to "uair?" which means the message makes more sense as a
	  "BACHOIR".

2003-11-25 13:38  kps

	* triail, triail.err: Disamb. "s" sufficiently well to catch no
	  lenition after "s"="six"

2003-11-24 22:56  kps

	* po/nl.po: Nearly complete Dutch translation, thanks to Anneke
	  Bart, <barta@slu.edu>.

2003-11-24 08:12  kps

	* po/ro.po: New Romanian translation, thanks to Eugen Hoanca
	  <eugenh@urban-grafx.ro>.

2003-11-12 11:22  kps

	* rialacha-ga.in: Couple of tags added for efficiency, plus
	  addition of the "NEAMHCHOIT" rule.

2003-11-11 22:49  kps

	* cuardach.c: Added requirement that the "lowered" word must always
	  be lowered in a rational way -- just first Letter, all LETTERS,
	  or just the lone capital after eclipsis "i nGaeilge" !

2003-11-11 21:19  kps

	* cuardach.c: Added addition of hyphen (nOibre ->  n-oibre, tOrd
	  -> t-ord) to lowercase conversion.

2003-11-11 21:09  kps

	* cuardach.c: Functionally, only change is that ALL uppercase
	  letters are made lowercase before secondary lookup; this way
	  bhFoirm, etc. will be recognized even without capped word
	  explicitly in IG -- unlike all the ispell/aspell agony.

	  Also separated out the lowercase and "code tidying" business for
	  cleanliness.

2003-11-11 20:44  kps

	* cuardach-ga.c: Added "rare word" markup.

2003-11-11 20:43  kps

	* cabhair.c: Smarts added when building focail-xx.bs files handling
	  "rare word" codes correctly -- in short, if there are alternate
	  codes, then throw away the 127 (==rare) code.

2003-11-11 20:42  kps

	* cuardach.c: (1) added stuff for "rare word" markup (2) no
	  "doubled word" markup if intervening punctuation

2003-11-04 11:59  kps

	* triail: (1) unstarred a couple of sentences in which errors are
	  now caught thanks to improved tagging.  (2) Fixed an error turned
	  up after "teanga" tagged correctly--	 "teanga dhothuigthe" (3)
	  changed "i r cuisneoir" to "i r gcuisneoir" so that the "inr"
	  is  reported instead of the missing eclipsis (newly caught b/c
	  "r" is now  correctly disambiguated) (4) added sentences for
	  doubled words and for the improved "abairti"

2003-11-03 14:42  kps

	* cuardach.c: Added qsort of the inflection codes when found as
	  UPPERCASE and lowercase.  Also strip repeats in this case.
	  changed to "unsigned char" for all instances of inflection codes.

2003-11-02 22:18  kps

	* aonchiall.meta.sed, gin.meta.sed: Change to aonchiall.meta
	  supporting new <Z> syntax, also added two macros to gin.meta.sed
	  for composing rules for this.

2003-11-02 22:17  kps

	* aonchiall-ga.in: First cut at adding Brill transformation-style
	  rules in template form.

2003-11-02 14:10  kps

	* po/de.po: New "repeated word" message.

2003-11-02 14:09  kps

	* Makefile.in: Added giorr to dist, etc.

	  Couple of maintainer level changes: benchmark file format and
	  stress testing.

2003-11-02 14:08  kps

	* gr.in: Added "giorr" improvement, escape_punc, to "abairti"
	  module.

2003-10-31 22:24  kps

	* giorr: New script containing abbreviations used by abairti.

2003-10-31 15:27  kps

	* stress.err, stress.out: Changes to German translation by Karl
	  Eichwalder.  Lots of changes to aonchiall => improved tagging in
	  XML.

2003-10-31 10:49  kps

	* cabhair.c: Small bug in construction of focail-xx.bs which meant
	  that the last grammar code of last word was not being found
	  (each line of *.bs technically needs to end in the "delimiter"
	  character -- here a newline)

2003-10-31 10:45  kps

	* Makefile.in, gr.in: Changes for "debug" and "benchmark" developer
	  targets.  Also a slight change in construction of focail-ga.bs
	  which eliminates proper name markup on words from .biobla which
	  are already in the main IG database (since these markups tended
	  to be redundant or concealed primary sense).

2003-10-31 10:43  kps

	* eagar.html: Typo corrected, thanks to Martin Gregory who was
	  trying to follow the incorrect instructions for mapping vim
	  hotkeys.

2003-10-31 09:43  kps

	* debug.sed: Script used by Makefile "debug" target, changes "gr"
	  to keep intermediate output and benchmark info.

2003-10-30 15:09  kps

	* cuardach.c: (1) strip repeated grammar codes when an upper case
	  word is found in db AND its lowercase is there too  (e.g. "Bord",
	  "Stt", etc.) (2) "no grammar codes" error moved so that
	  offending word prints out too

2003-10-29 22:16  kps

	* gr.in: Added second pass through aonchiall and the "repeated
	  word" message to loganu.

2003-10-29 22:15  kps

	* eisceacht-en.in: "that" -- only exception I know in English.

2003-10-29 22:15  kps

	* eisceacht-ga.in: Exceptions to repeated word rule.

2003-10-29 22:15  kps

	* cuardach.c: Added stuff for marking up repeated words.

2003-10-29 22:14  kps

	* aonchiall-ga.in: Major changes, additions, reorganization based
	  on output of Brill algorithm.   Now running aonchiall-xx.pl twice
	  to "bootstrap".   Also moved some "last resort" taggings to
	  bottom in the hope that bootstrap will work for   these, avoiding
	  the last resort, with   just one pass through aonchiall.

2003-10-29 22:12  kps

	* Makefile.in: Moved "vim" up to user target, utfcheck down to
	  maintainer.  Added rigeail to build process.	Added emacs/vim to
	  dist.

2003-10-29 22:11  kps

	* rigeail: Script for fixing up aonchiall-xx.pl backreferences
	  depending on the complexity of the macros expanded.

2003-10-29 22:10  kps

	* po/: ga.po, gramadoir.pot: New message for "Repeated word."

2003-10-25 15:11  kps

	* eagar.html: Edits to the emacs section and new Vim section added.

2003-10-25 14:18  kps

	* gramadoir.vim: No more "start Gramadir" hot key-- now just use
	  the same as the one to step through errors.  Also made the
	  function calls silent and print a "fan go fill..." instead upon
	  loading.

2003-10-24 15:05  kps

	* gramadoir.vim: First cut at a Vim interface.

2003-10-23 20:50  kps

	* po/de.po: Formatted changes after msgfmt.

2003-10-23 20:38  kps

	* gramadoir.el: Emacs interface contributed by Martin Gregory
	  <Martin.Gregory@eur.sas.com>.

2003-10-23 20:36  kps

	* cuidiu.html, eagar.html, foirm.html, index.html, iompar.html,
	  samptemplate.html, sios.html, sonrai.html: Trivial changes
	  reflecting addition of eagar.html and supporting rhetoric.

2003-10-23 20:21  kps

	* eagar.html: New file describing the emacs interface.

2003-10-23 20:04  kps

	* cuidiu.html: Link to Montreal.

2003-10-22 16:28  kps

	* cuardach-ga.c: Missing close quote in attribute -- how'd I miss
	  that?

2003-10-22 10:25  kps

	* iompar.html: Tidying.

2003-10-22 10:25  kps

	* eile-en.bs: Built by collating English and American word lists
	  from ispell.	A bit of hand-editing at the end.

2003-10-22 10:23  kps

	* aonchiall-en.in, eisceacht-en.in, rialacha-en.in: Trivial en
	  implementation.

2003-10-21 23:42  kps

	* po/de.po: Modifications to Martin Gregory's original German
	  translation; thanks to Karl Eichwalder of the GNU Translation
	  Project.

2003-10-21 16:27  kps

	* po/ga.po: Rejected from Montreal because revision date was before
	  POT creation date!

2003-10-21 11:03  kps

	* po/: de.po, ga.po, gramadoir.pot: Time stamps.

2003-10-21 11:02  kps

	* ponc.in.l: Generalize so that attributes are allowed on RHS of
	  rules in aonchiall.

2003-10-21 11:01  kps

	* configure.ac, index.html, sios.html, stress.out: Version number
	  bumped.

2003-10-21 10:16  kps

	* Makefile.in: don't need m4 stuff for configure.

2003-10-21 10:15  kps

	* po/: de.po, ga.po, gramadoir.pot: New timestamps, new line
	  numbers only.

2003-10-21 10:14  kps

	* sonrai.html: diutf description corrected/updated.

2003-10-21 10:13  kps

	* sios.html: Added link to NLP registry.

2003-10-21 10:13  kps

	* iompar.html: Rewrite based on existence of, and experience with,
	  English port.

2003-10-21 10:12  kps

	* index.html: Builds on Mac OS X now.

2003-10-21 10:10  kps

	* cuidiu.html: Mention GNU TP site.

2003-10-21 09:50  kps

	* gr.in: Added support for specifying internal and boundary chars
	  for each language.  Only subtle thing is the addition of the
	  LC_COLLATE=C before the appropriate "sed" command in "abairti" so
	  that "A-Z" works as intended	(w/ iso88591 locale, this range
	  includes accented letters).

2003-10-21 09:47  kps

	* cuardach.c: Made "markup" recursive since before it wasn't
	  dealing with sequences of more than one word not separated by
	  whitespace: "Kevin+Scannell" would markup as
	  "<Y>Kevin</Y>+<c>Scannell</c>".  Now the tail gets sent to markup
	  too.	Wasn't handling English possessives correctly because of
	  this, or "and/or" in the GPL.

2003-10-21 09:45  kps

	* configure.ac: New variable for the "cuardach-xx.o"'s; used as as
	  dependencies in Makefile.

2003-10-21 09:43  kps

	* TEANGACHA: Added English, and also added specification of
	  internal and boundary characters.

2003-10-21 09:43  kps

	* Makefile.in: Generalized targets to handle "cuardach-xx",
	  "focail-xx", "eile-xx", and the "farc" stuff.

2003-10-21 09:39  kps

	* stress, stress.out: (1) Because of new, language-dependent
	  tokenizing, English check of GPL now recognizes possessives,
	  "and/or" correctly.  (2) Added new check of non-ascii chars w/
	  teanga=en.

2003-10-20 22:36  kps

	* stress, stress.out: Added English language test -- spellcheck
	  head of GPL!

2003-10-17 16:15  kps

	* cuardach.c, farc.c.in: Moved the "byte_to_markup" function out of
	  cuardach into this template C file which is generated w/ some sed
	  (at configure time) from the TEANGACHA data file.

2003-10-17 16:14  kps

	* po/LINGUAS: German translation thanks to Martin Gregory.

2003-10-17 16:13  kps

	* po/Makefile.in.in: A hack to avoid repeating the "not supported"
	  message that it is gr.in AND farc.c.in

2003-10-17 16:13  kps

	* po/POTFILES.in: Moved a couple of messages from cuardach.c to
	  farc.c.in

2003-10-17 16:12  kps

	* po/de.po: Formatting changes, and trivial changes because of pot
	  update.

2003-10-17 16:12  kps

	* po/ga.po: Changes corresponding to pot changes, plus: misspelled
	  "parsil" and switched to "brabhsla" from "brabhsaer"

2003-10-17 16:10  kps

	* po/gramadoir.pot: Generalized "corrupted" message.  Moved a
	  couple to "farc.c.in".

2003-10-15 23:59  kps

	* gr.in: New command line argument to "cuardach" call.

2003-10-15 23:58  kps

	* cuardach-en.c, cuardach-ga.c: The new byte_to_markup c files.
	  "ga" is just the old code from cuardach.c, and the "en" one is
	  trivial, always return <U></U> markup.

2003-10-15 23:57  kps

	* cuardach.c: (1) dynamically allocate *.bs (2) new command line
	  argument with language code, load    correct versions of *.bs
	  files, and do the byte_to_markup    depending on the language --
	  stored in cuardach-xx.c files.

2003-10-15 23:48  kps

	* stress: Changed name of database files => change to stress test
	  script.

2003-10-14 23:34  kps

	* stress, stress.err, stress.out: Added new test for when "iconv"
	  aborts b/c of unknown encoding.

2003-10-14 20:36  kps

	* Makefile.in: Added stresstest stuff.

2003-10-14 20:35  kps

	* gr.in: Restructured main pipe sequences to avoid repetition, no
	  change in fuctionality.

	  Also now redirect all error messages to stderr (some were going
	  to stdout)-- there was also a bug for two errors where a missing
	  "&" was causing messages to go to a file named "2"!

2003-10-14 20:26  kps

	* stress, stress.err, stress.out: Local development scripts for
	  testing, error-trapping.

2003-10-14 15:41  kps

	* cuardach.c: To avoid broken pipes, need to eat up stdin in
	  cuardach before exiting with nonzero status.

2003-10-14 15:40  kps

	* Makefile.in: Delete lastperl in clean.

2003-10-07 20:45  kps

	* po/de.po: German PO file added, thanks to Martin Gregory for the
	  translation.

2003-10-02 22:44  kps

	* cuardach.c: Embarrassing indexing bug in "cleanup".  Didn't
	  manifest itself on the Gentoo systems, maybe b/c they have lots
	  of memory, but when ported to Rice, got an incredibly weird bug
	  that the output of "cuardach" would go to stdout but couldn't be
	  redirected or piped anywhere else (in particular, not to the rest
	  of "gr" or even to an inserted "tee").

2003-10-02 21:36  kps

	* TEANGACHA: Comment tweak.

2003-10-02 21:36  kps

	* gr.in:
	  Added quotes to "${1}" filename arg throughout (and other places
	  too) so that filenames containing spaces are handled correctly.
	  Also, because of this, can't assume `cat "$1"' is ok anymore if
	  $1 is empty, so restructured the three "ACTION" functions as pure
	  filters.

2003-10-01 22:22  kps

	* Makefile.in: Changes to "make installcheck", etc., using
	  LC_MESSAGES=C now for triail.err.

2003-10-01 22:21  kps

	* triail.err: Since we can't assume anything about LC_MESSAGES,
	  LANG, etc. or even what locales are installed on target systems,
	  the "make installcheck" now uses LC_MESSAGES=C when generating
	  output and compares with the (new) English language triail.err
	  file.

2003-10-01 22:10  kps

	* Makefile.in: Taking triail.utf out of distribution (just added it
	  earlier today) since older versions of "iconv" seem not to
	  support utf conversion uniformly, and shouldn't have "make
	  installcheck" complain because of this if the rest of the install
	  is OK!   Also changing language of messages from "make
	  installcheck" from Irish to English to help maintainers.

2003-10-01 22:01  kps

	* Makefile.in: Apparently "old" sed's don't like "\n" in the
	  replacement string (though tests fine when used on the LHS, e.g.
	  at end of gin.meta.sed).  So this is a hack getting around that
	  when building the .pl's.

2003-10-01 22:00  kps

	* po/: ga.po, gramadoir.pot: Trivial changes after rebuild.

2003-10-01 15:38  kps

	* TEANGACHA: New file for specifying all available languages.  It
	  is part of dist and installed in share.

2003-10-01 15:37  kps

	* po/ga.po: getopts strings translated.

2003-10-01 15:37  kps

	* po/gramadoir.pot: New strings added for "getopts" stuff.

2003-10-01 15:37  kps

	* index.html: --ionchod description added.

2003-10-01 15:36  kps

	* grweb: Since "httpd" user has no LC_MESSAGES default, need to set
	  it explicitly here so that web interface comes out in Irish.
	  Eventually should add a radio button to the web page or sumpin.

2003-10-01 15:35  kps

	* gr.in: (1) stripped out comment describing command line opts
	  since it is useless and a pain.  (2) Added all of the stuff for
	  the --ionchod command line option;	 also read "native"
	  encoding for the language specified at runtime,     if different,
	  then "diutf" now uses iconv.	(3) tidied up "getopt" errors a
	  bit.

2003-10-01 15:33  kps

	* configure.ac: Syntax change -- now includes character encoding
	  too.

2003-10-01 15:33  kps

	* Makefile.in: (1) TEANGACHA in dist and also installed in
	  pkgdatadir so that	the char encoding names can be read at run
	  time.  (2) Added a utf version of triail to the dist and the
	  make installcheck target.

2003-10-01 15:31  kps

	* ABOUT-NLS: tweak.

2003-09-28 13:40  kps

	* cuardach.c, gr.in, po/ga.po, po/gramadoir.pot: Made the program
	  title "An Gramadir" part of the pot file since one can't assume
	  the Latin-1 character will appear correctly for everyone.

2003-09-18 23:57  kps

	* po/Makefile.in.in, configure.ac, gr.in: Changes for disabling NLS
	  with shell script.  Handled fine for C/bison stuff by the -D
	  flags passed and gettext.h, but the shell scripts used to have
	  explicit "gettext -s" calls.	These are now SUBST'd with "echo"'s
	  if NLS is disabled in configure.

2003-09-18 23:54  kps

	* cuardach.c, ponc.in.y: Even when NLS is off, need LC_MESSAGES
	  variable available, so include <locale.h>.   Found this on
	  axa.slu.edu.

2003-09-18 23:22  kps

	* Makefile.in: Adding this ChangeLog to dist target.

2003-09-18 21:29  kps

	* Makefile.in, cuardach.c, ponc.in.y: Changes b/c of "gettext.h"
	  vs. <libintl.h>

2003-09-18 20:29  kps

	* gettext.h: Got this out of the awk dist (randomly).  See gettext
	  manual section 12.4 for details-- basically handles conditional
	  include of libintl if NLS is not disabled and headers are
	  installed.

2003-09-18 19:58  kps

	* po/: LINGUAS, Makefile.in.in, Makevars, POTFILES.in, Rules-quot,
	  boldquot.sed, en@boldquot.header, en@quot.header, ga.po,
	  gramadoir.pot, insert-header.sin, quot.sed, remove-potcdate.sin:
	  Adding all intl stuff to version control.

2003-09-18 19:51  kps

	* gr.in, index.html: Added description of the --teanga flag to web
	  page and online help.

2003-09-18 19:51  kps

	* Makefile.in: Forgot to include TEANGACHA in the tarball.

2003-09-18 12:08  kps

	* aonchiall.in, eisceacht.in, rialacha.in: Renamed with "-ga"
	  inserted before "."

2003-09-18 12:07  kps

	* eisceacht-ga.in, gr.in: Trivial name changes in comments.

2003-09-18 12:07  kps

	* iompar.html: Name changes adding "-ga" or "-xx".  Also note that
	  I'm now using gettext in instructions about translating the UI.

2003-09-18 12:06  kps

	* Makefile.in:
	  Hopefully final changes reflecting name changes (in dist target,
	  etc.)

2003-09-18 12:05  kps

	* cuidiu.html, sonrai.html: Just adding "-ga" or "-xx" to
	  filenames.

2003-09-18 11:51  kps

	* aonchiall.meta.sed, eisceacht.meta.sed, rialacha.meta.sed: Just
	  changed comments to reflect new names.

2003-09-18 11:49  kps

	* aonchiall-ga.in, eisceacht-ga.in, rialacha-ga.in: Just changed
	  names of these three files, adding the ISO-639 language code
	  since it will be used when new languages are added.

2003-09-18 11:27  kps

	* Makefile.in: Lots of changes: (1) Internationalize poncin
	  executables=> need $(DEFS) passed for LOCALE_DIR...  (2)
	  Recursively call make targets in po in various places.  (3) Add
	  scripts to tarball (4) Rename perl scripts adding language code
	  (5) Changed make scheme so that it will create perl scripts for
	  all	  languages automatically with correct dependencies  (using
	  .in.pl suffix )

2003-09-18 11:23  kps

	* configure.ac: Changes allowing multiple languages in Makefile.

2003-09-18 08:42  kps

	* gr.in:
	  Internationalized grammatical errors themselves.

2003-09-18 08:41  kps

	* triail.err:
	  Changed error message slightly since it became apparent when
	  translating to English that it wasn't entirely clear.

2003-09-18 08:24  kps

	* gr.in: Added --teanga=xx option.

2003-09-17 20:35  kps

	* gr.in:
	  This works now locally AND on patel2 (which for some reason
	  didn't like the $"" business AT ALL -- works completely fine with
	  "gettext -s" though).

	  Had to mess with the po/Makefile.in.in so the gramadoir.pot
	  target works (since it extracts the strings marked with $"" and
	  not the gettext -s ones!)

2003-09-17 20:33  kps

	* ponc.in.y:
	  All the necessary stuff for internationalizing the .in file
	  checkers.   Note that the msgid_plural/ngettext business is
	  required for some of the messages.

	  Also, some of the output (e.g. "parse error") comes directly from
	  flex or bison so if those aren't localized, the end user will see
	  English.

2003-09-17 19:40  kps

	* ABOUT-NLS:
	  Generic GNU file discusses intl/ directory a lot and the fact
	  that installers needn't have GNU gettext which isn't true in our
	  case because of the bash gettext calls...

	  I pared things down, also stripping out the outdated matrix which
	  doesn't show ga locale!

2003-09-17 15:25  kps

	* gr.in:
	  This version works on borel.

	  Needed to export the TEXTDOMAIN stuff to get "gettext -s" to
	  work.

	  Also changed so that $0 is used to get the executable name "gr".

	  Had to change a few $"" to gettext -s, essentially inexplicably.

2003-09-17 10:03  kps

	* cuardach.c: Bug fix -- discovered that it is not sufficient to
	  set LC_MESSAGES if you want accented characters to appear --
	  thanks to a post by Tomohiro Kubota to lists.debian.org for this
	  explanation.

2003-09-16 23:35  kps

	* mkinstalldirs: In adding gettext scripts, I noticed that this
	  script was never added to version control (even though unlikely
	  I'll change it!)

2003-09-16 23:34  kps

	* config.guess, config.rpath, config.sub: Three scripts downloaded
	  from ftp.gnu.org (or taken right from hello distribution).  Only
	  the 2nd one is claimed necessary by gettext manual if one isn't
	  using intl/  (I'm not) but configure complains if 1st/3rd aren't
	  present.

2003-09-16 23:32  kps

	* aclocal.m4: Just a cat of a few m4 files out of the gettext
	  distribution.  See the GNU gettext manual, section 13? Since not
	  using intl/, don't need all of them!

2003-09-16 23:31  kps

	* ABOUT-NLS:
	  Now part of the disttribution per the coding standards for
	  internationalized software.

2003-09-16 23:30  kps

	* Makefile.in: First pass at adding gettext stuff.  localedir now
	  works, and if I install an .mo file on my own, the package is
	  indeed localized.

	  Need to fix the dist target, install target, etc.  to use the po/
	  stuff.

2003-09-16 23:28  kps

	* cuardach.c, configure.ac: Gettextized!

2003-09-16 23:28  kps

	* foirm.html:
	  Added "." in grweb, so no longer really require punctuation, but
	  added this sentence anyway.

2003-09-16 23:27  kps

	* gr.in:
	  A ton of changes but all for i18n, gettext stuff.

2003-09-16 23:26  kps

	* grweb: Add period at end of input to avoid the most common
	  confusion with the web app!

2003-09-16 23:25  kps

	* index.html: Changed translation link to my new local page.

2003-09-16 23:25  kps

	* sios.html: Link to gentoo package.

2003-08-25 21:33  kps

	* Makefile.in: (1) changed make of the .pl targets to first
	  separate out the stuff after the colon's in *.in onto new lines
	  -- then pass to gin.meta.sed (which was correspondingly
	  rewritten).  (2) minor bug in "make triail.err" -- need to use
	  flag --iomlan so answer will be uniform across systems (now
	  matches the "make installcheck" flag which correctly uses
	  --iomlan)

2003-08-25 21:31  kps

	* gin.meta.sed: Now when this script starts, the stuff after the
	  colon is now on the following line (which starts with a colon).
	  So we now only process lines NOT beginning with # or :.
	  Simplifies dealing with words -- now they are uniformly
	  surrounded by one space on each side.

	  At the end, use N to rejoin.

2003-08-25 21:29  kps

	* gr.in: Trivial formatting changes; PLUS no longer need to strip
	  out underscores from BACHOIR.

2003-08-25 21:28  kps

	* ponc.in.l: No trailing whitespace allowed anymore.  Also, spaces
	  instead of underscores in MESSAGEARG's.

2003-08-25 21:27  kps

	* rialacha.in: (1) no underscores in BACHOIR's anymore.  (2) no
	  trailing whitespace allowed anymore -- caught 4 lines with it (3)
	  uncommented (and added) adjective lenition rules

2003-08-25 21:26  kps

	* rialacha.meta.sed: Since we're now allowing whitespace after
	  colon, generalize what is matched in these two rules.

2003-08-25 21:25  kps

	* triail, triail.err: Corrected an error found when adjective
	  lenition rules re-inserted!

2003-08-25 15:28  kps

	* configure.ac: Added -r1 to version to distinguish output
	  generated by webapp from the actual 0.2.  I'll decide as I go
	  when to increment this  (probably much more often than actual
	  tarball releases?)

2003-08-25 15:27  kps

	* gr.cgi: Terrible mistake caused plain html to be rendered on
	  Mozilla/lynx (though not IE, probably since they don't go by
	  standards).

	  Fixed just after sending out url to gaeilge-a, etc.

2003-08-24 21:04  kps

	* cuardach.c: Took out ctype/locale stuff since one user was having
	  problems and needed to install locale.  Might as well use
	  assumptions about the data!

2003-08-24 21:02  kps

	* cuidiu.html, index.html, iompar.html, samptemplate.html,
	  sios.html, sonrai.html: Changes for addition of foirm.html; some
	  rhetoric added.

2003-08-24 14:09  kps

	* Makefile.in: Added Neamhshuim to "make installweb" -- note at
	  this stage I haven't added it to "make dist"!  and certainly not
	  to "make install"!

2003-08-24 14:09  kps

	* cuardach.c: cuardach now looks for a global Neamhshuim file if
	  there isn't one in ${HOME}.	Also, added a check of return value
	  from getenv which caused a crash when run by "apache"!!!

2003-08-24 14:07  kps

	* gr.cgi: Tiny bug; wasn't causing problem, just want valid HTML
	  returned!

2003-08-24 14:07  kps

	* grweb: Added global Neamhshuim file for web app.

2003-08-24 13:13  kps

	* gr.in: Added to gr --help, descriptions of options in Irish.

2003-08-23 14:35  kps

	* Makefile.in: Additions to installweb which install foirm.html AND
	  ALSO the scripts in secret locations!  Even names of these files
	  should remain secret.

2003-08-23 14:34  kps

	* foirm.html: Form instruction tweaks.

2003-08-23 14:33  kps

	* gr.cgi: Added "Tainted" flag, requires the regexp check on
	  "ionchur".  Since single quotes are needed in Irish, escape 'em.

2003-08-23 14:32  kps

	* gr.in: Two hacks:  htmlversion function (requires --html flag
	  before --version flag)

	  and strip "\015"'s which were coming in from web input form.

2003-08-23 14:31  kps

	* grweb: Added logging line.

2003-08-23 12:42  kps

	* gr.cgi, grweb: Working version of web form, but insecure.

2003-08-23 12:40  kps

	* foirm.html: Tweaked html form interface.

2003-08-22 20:28  kps

	* foirm.html: First cut at the HTML form for a web interface.

2003-08-15 14:17  kps

	* Makefile.in: Only a change to my personal "neamhshuim" target,
	  adding strings to ignore from my *.po GNU translation files.

2003-08-15 14:17  kps

	* sonrai.html: Long <pre>'d line was causing CSS nav column to go
	  to bottom on thin browsers.

2003-08-15 14:08  kps

	* gr.in: Stripped out all backslashes and any stray <,>  left over
	  after stripping SGML-like tags.  These were causing crashes when
	  spellchecking heavy-C-syntax stuff like the *.po files for GNU.

2003-08-06 10:00  kps

	* cuardach.c: Two changes: (1) lookup "lowered" in ignore file too
	  (already looking it up    in the regular dictionary and the
	  alternate dict) (2) changed the meaning of "lowered" so that ALL
	  uppercases are lowered;    this way ROGHA, etc. won't be
	  reported.   Currently conflicts    with ispell definition though
	  since GnthFhocal won't be reported either.

2003-08-01 16:44  kps

	* cabhair.c, cuardach.c: Ran "indent" on them both.  That's all.

2003-07-30 11:22  kps

	* configure.ac, cuardach.c, index.html, sios.html: Trivial changes
	  to version number, file size in sios.html and word count in
	  cuardach.c

2003-07-29 21:37  kps

	* cuardach.c: Small change so that space isn't printed after a
	  prematurely terminated token (they're cut off at len==511
	  arbitrarily).

2003-07-29 21:28  kps

	* cuardach.c: Hardened against buffer overflow attack.	I had %512s
	  in there; changed to %511s since I need the NUL char at the end
	  (even though tests indicate that scanf puts one there anyway).

	  Also, there was a crash when a length 511 string with a <c> but
	  no </c> was passed to "markup".   Error check on the "strstr"
	  call for "</c>" now too.

2003-07-29 16:23  kps

	* sonrai.html: Minor boo-boo in description of the meta-language.

2003-07-29 16:00  kps

	* aonchiall.in, triail, triail.err: Disambiguation of "leis"
	  (partial).  Allows "leis an farraige" to be caught in "triail".

2003-07-29 15:44  kps

	* aonchiall.in, rialacha.in: p. 84 and p. 132 of CB94; added a
	  whole mess of preposition+particle rules (indicating when they
	  should combine and indicating eclipsis after prep+r forms)

2003-07-29 15:43  kps

	* Makefile.in: Changed installcheck to not depend on line numbers
	  so that I get a meaningful "diff" if I insert a line early in
	  "triail".

2003-07-29 15:40  kps

	* triail, triail.err: New lines for the eclipsis rules prep+r,
	  etc.	Also some combos prep+a/ar CB p.132.

2003-07-29 10:57  kps

	* triail, triail.err: Changes for new p.84 rules.

2003-07-28 23:03  kps

	* triail.err, cuidiu.html, gr.in: Misspelled shave-you.

2003-07-28 22:51  kps

	* triail.err: Just added the "non-standard" error message near the
	  top (so diff is a mess --only line numbers should be changing
	  though)

2003-07-28 22:50  kps

	* triail: New sentence demonstrating the non-standard error
	  message.

2003-07-28 22:50  kps

	* cuardach.c: Small bug fix -- now also do replacement lookup on
	  lowercase version when the token in the text is uppercase.

2003-07-28 22:20  kps

	* cuardach.c: Bug caused by insufficient buffer size for "repl"
	  passed to replacementlookup.	 Also reversed order so that the
	  ignore lookup happens before the replacement markup so that a
	  user can put favorite non-standard forms into the ignore file.
	  Re-enumerated "tail" pointer for no reason.

2003-07-28 16:07  kps

	* Makefile.in, cabhair.c: Changes which meld together "replacement"
	  strings into a single comma-separated entry.	 Executable
	  "cabhair" is now used for developer creation of both focail.bs
	  and eile.bs ("cabhair 0", "cabhair 1" respectively)

2003-07-28 16:05  kps

	* cuardach.c: Fixed a bug caused by fixed-size allocation of the
	  athfhocal replacement string.  Using malloc now and have "free"
	  in cleanup function called from main before return.  Added a
	  couple of comments too.

2003-07-28 12:05  kps

	* cuardach.c: Added "replacement file" functionality.	Some error
	  markup actually occurs in this file now!

2003-07-28 12:04  kps

	* gr.in: Added UI line for "CAIGHDEAN".

	  Also moved to "case" statement in dealing with command line opts.

2003-07-28 12:03  kps

	* ponc.in.l: Added macro "CAIGHDEAN" even though not currently used
	  in the actual rialacha.in.   It is used in cuardach.c!  Only
	  error markup which occurs outside of rialacha at the moment.

2003-07-28 12:03  kps

	* iompar.html: Added sentence about constructing "eile.bs".

2003-07-28 12:02  kps

	* Makefile.in: Added developer target for building MY
	  ~/.neamhshuim.

	  Also added the lines for including eile.bs in the distribution.

	  Redid the meaning of -DBSONRAI to expand to just the directory
	  name.

2003-07-26 21:29  kps

	* Makefile.in: Now that .neamhshuim is added to default system,
	  want the make installcheck to be independent of any given user's
	  file, so added --iomlan flag to this make target.

2003-07-26 21:28  kps

	* cuardach.c: Mostly stuff for the "ignore" file, including a
	  command line argument added indicating whether or not to use it.

	  Other minor tidying: made the main token scanf safe with a %512s,
	  and added some #define's for language or OS specific stuff (max
	  word len, etc.)

2003-07-26 21:26  kps

	* gr.in: Changes because of --iomlan command line option and now
	  executable "cuardach" takes an option "ignore" or "noignore"
	  also.

2003-07-26 21:25  kps

	* index.html: Added info about creating a .neamhshuim and added the
	  --iomlan command line option description.

2003-07-26 14:23  kps

	* triail.err: Change in triail.err reflects debugging of the
	  "multiple misspellings in once sentence troubles"

2003-07-26 14:21  kps

	* ponc.in.l, ponc.in.y: Lex tokenizer and bison grammar for the .in
	  files.  Forgot to add to cvs before the 0.1 release (they were
	  really in there though!)

2003-07-26 14:20  kps

	* gr.in, rialacha.in, rialacha.meta.sed: Tiny changes which
	  disallow multiple misspelling errors reported for the same
	  sentence.  Also fixes the bug that caused every other word in an
	  English sentence to be marked as misspelled (since the
	  misspelling rule was being applied globally before the </E>'s
	  were fixed up.)  moot point now.

2003-07-26 14:17  kps

	* iompar.html: Added the caveat emptor that this is just a regexp
	  checker.

	  Also corrected the assertion that a plain word list is
	  distributed with aspell (closer with ispell actually)

2003-07-22 15:33  kps

	* rialacha.in: Subtle bug showing up in triail output.	"bhur dh
	  x" was giving "missing eclipsis" error as desired, but only
	  highlighting "bhur dh" since the eclipsis rule w/o "dh" is
	  being applied first.

	  Revesed these and the ones for "r"  (though no error their since
	  this requires disambiguation).

	  Note that triail.err is unchanged since the highlighting doesn't
	  appear in that file!

2003-07-21 16:58  kps

	* cuardach.c, iompar.html: Added translator help to the strings in
	  cuardach.c and noted the existence of these strings on web page.

2003-07-21 16:11  kps

	* gr.in: Added comments for translators+those who might want to
	  port to checkers for other languages.

2003-07-20 14:01  kps

	* index.html, sonrai.html: Added release history and fixed some
	  typos.

2003-07-20 13:31  kps

	* gr.in: Added error for unknown option.

2003-07-20 13:17  kps

	* gr.in: Silly bug in gr error message.

2003-07-19 23:16  kps

	* configure.ac, gr.in:
	  Changes allowing configure script to search for a good regexp
	  highlighter (egrep --color or glark) and upon failure use vanilla
	  egrep.

2003-07-19 15:36  kps

	* Makefile.in: Using autoconf variables more consistently now.
	  (Lex file names, DEFS, etc.)	Also had incorrect VERSION variable
	  in there for no reason.

2003-07-19 15:35  kps

	* cabhair.c, cuardach.c: Using autoconf STDC_HEADERS variable now.
	  Also had some non-portable C++ type stuff in cabhair.c (comment
	  and a var. declared in the middle of a fn)

2003-07-19 15:34  kps

	* gr.in: Fixed program name.  Using @PERL@ autoconf variable.

2003-07-18 13:46  kps

	* index.html, iompar.html: Trivial changes to web pages.

2003-07-18 13:46  kps

	* sios.html: New version number.

2003-07-18 13:43  kps

	* configure.ac: Updated version number to 0.1

2003-07-18 13:40  kps

	* cabhair.c, cuardach.c: Added copyright notices, updated word
	  count in cuardach.

2003-07-18 13:40  kps

	* Makefile.in, aonchiall.in, eisceacht.in, gr.in, hilite.awk,
	  rialacha.in: Added Copyright and pointer to COPYING.

2003-07-18 13:39  kps

	* aonchiall.meta.sed, eisceacht.meta.sed, gin.meta.sed,
	  rialacha.meta.sed: Added Copyright notice, pointer to COPYING.

2003-07-18 13:20  kps

	* gin.meta.sed: Two new macros.

2003-07-18 13:20  kps

	* eisceacht.in: Some false positives discovered in TB corpus.

2003-07-18 13:19  kps

	* rialacha.in, triail.err: Eliminated some false positives
	  discovered while checking TB '99.  Since I disambiguated "ba"
	  recently, I now know that the lenition rule is irrelevant for
	  prepositions (quite common: "ba den ...", etc.)

	  The rules about File, Naomh	not leniting are now restricted to
	  capitals.

	  switched the order of "le + vowel" and "le + an" checks so that
	  latter will be applied!   This changes one line of triail.err!

2003-07-18 12:52  kps

	* aonchiall.in, triail, triail.err: Added default disambiguation of
	  capital "San" to <S> when it isn't followed by a capital.  Added
	  new line to triail to verify this works.

2003-07-18 12:36  kps

	* aonchiall.meta.sed: Allowing space after rule in aonchiall.in.
	  Substitutions doesn't need to be global since it starts and ends
	  with ^,$

2003-07-18 12:35  kps

	* aonchiall.in:
	  Added rules for "ba", "faoin", "gur", "" (lowercase), "San"
	  (uppercase).

2003-07-18 12:18  kps

	* Makefile.in: Fixed a bug in generation of focail.bs b/c I forgot
	  that ${HOME}/.biobla contains some ispell affixes which need to
	  be expanded!

2003-07-18 12:17  kps

	* rialacha.in: Liberalization of "mo" rule to catch "mo aitheanta"
	  in triail.

2003-07-18 12:16  kps

	* triail, triail.err:
	  DONE!   triail has lines for virtually every rule, and except for
	  some hard disambiguations (with asterisks) they all are reported
	  as errors.

2003-07-18 12:01  kps

	* triail, triail.err: Couple of hours of hacking to improve rules,
	  change sentences to avoid ambiguities.  All but a handful of
	  lines in "triail" show up as errors now, and the ones which don't
	  are marked with an asterisk and noted in "Tascanna" as important
	  disambiguations.

2003-07-18 09:09  kps

	* triail.err: First cut at errors in new triail.  About to debug.

2003-07-18 09:08  kps

	* triail:
	  Last pass at additions to triail.  Not debugged yet.

2003-07-18 07:58  kps

	* sonrai.html: Comments regarding "meta" added to sonrai.html.

2003-07-17 22:46  kps

	* gr.in:
	  Major restructuring to allow more command line options.  For now,
	  implemented --xml and --litriu.

2003-07-17 22:45  kps

	* cuidiu.html, index.html, sonrai.html:
	  Changes to web pages because of new command line options --xml,
	  --litriu

2003-07-17 20:28  kps

	* Makefile.in, index.html: Compliance with GNU coding standards.
	  make check is supposed to work pre-install, which I can't do w/o
	  restructuring things.  Also, make installcheck shouldn't assume
	  "gr" is in the path.

2003-07-17 19:53  kps

	* eisceacht.in: Some additions and some generalizations based on
	  testing of the OF81 corpus.

2003-07-17 19:52  kps

	* Makefile.in:
	  A few minor changes, some tidying to the clean targets to bring
	  'em more in line with GNU standards, and added targets for the
	  Makefile itself, gr, config.status, etc. so the autoconf tools
	  will be called when necessary.

2003-07-17 19:50  kps

	* gr.in:
	  Added bug-report macro from autoconf.  Added AWK macro in places
	  where it was missing.

	  Also fixed a subtle portability bug -- old sed versions don't
	  recognize \t as tab in patterns.  Added a shell variable TAB and
	  now I use that.

2003-07-17 15:40  kps

	* aonchiall.meta.sed, gin.meta.sed:
	  Small bug fix -- checked for any tags [A-Z] in a couple places in
	  the meta.sed scripts -- changed to A-DF-Z so that error tags are
	  distinguishable from grammatical tags.   Otherwise we'd get
	  nested errors and invalid XML.

2003-07-17 12:11  kps

	* gr.in:
	  Made to work with autoconf/configure script now.

	  Couple of other minor bugs: (1) strip all "&"'s from the input
	  file so that the final XML stream   validates OK.  See abairti.
	  (2) POSIX sed doesn't allow whitespace after a negated pattern
	  /.*/!    Discovered while porting to axa.

2003-07-17 12:09  kps

	* cuardach.c:
	  Added a new locale choice after building on "axa".  No need for
	  the LC_COLLATE variable.

2003-07-17 12:08  kps

	* Makefile.in: Some serious changes to work with new autoconf
	  scheme.  (1) never run program out of local directory now, so
	  don't worry about permissions (2) Fixed the install target to fit
	  the GNU coding standards better --	use mkinstalldirs script
	  and include full filenames as 2nd args     (not just dirs) (3)
	  some tidying since I now consider triail.err as part of the
	  software configuration (even though it can be built on the fly)
	  (4) added new files to the tarball.  I'd forgotten the ponc.in.*
	  stuff too.

2003-07-17 12:05  kps

	* index.html:
	  New installation instructions including configure step.

2003-07-17 11:53  kps

	* triail.err: Decided to keep this under source control to
	  formalize the "checking" process.

2003-07-17 10:28  kps

	* install-sh: As I understand it, autoconf requires that this file
	  be present in case an old system has a non-compliant or missing
	  install.

2003-07-17 10:22  kps

	* Makefile.in, configure.ac: First try and using autotools.   Seems
	  to be building OK.

2003-07-17 10:03  kps

	* Makefile: Using autoconf now.

2003-07-16 16:03  kps

	* Makefile: Added stuff for the bison/flex grammar.  Cleaned up the
	  "clean" targets.  Separated out "developer" targets.

2003-07-16 16:01  kps

	* index.html: Mention "bison" and "flex" under System Requirements.

2003-07-16 16:00  kps

	* iompar.html: Minor changes; noted buffer size in cuardach.c

2003-07-16 15:59  kps

	* sonrai.html: Details on aonchiall added.

2003-07-16 15:59  kps

	* cuidiu.html: Now that grammar is essentially done, I've added
	  descriptions to the "contributing" page.

2003-07-16 15:58  kps

	* triail: Sample sentences actually added Tuesday 7/15/03.

2003-07-15 14:36  kps

	* index.html:
	  Tried and failed building on Mac OS X so I added these lines to
	  System Requirements!

2003-07-15 14:36  kps

	* Makefile:
	  Tiny bug in "make dist" - discovered while testing build on other
	  machines.

2003-07-15 13:56  kps

	* cuardach.c: cuardach now outputs the XML header and
	  <teacs></teacs> tag at the top/bottom of every file.

2003-07-15 13:55  kps

	* sonrai.html:
	  Added lines about new file gramadoir.dtd

2003-07-15 13:55  kps

	* Makefile:
	  added new file gramadoir.dtd to dist.

2003-07-15 13:47  kps

	* gramadoir.dtd:
	  New file.   Currently validates pre-comheadan marked up text
	  stream (though it gives warnings because of	(1) use of "empty"
	  tags <S/> etc. inside the <Z></Z>	  even though they're
	  supposed to have "PCDATA"   (2) multiple uses of attribute values
	  "y" and "n"

2003-07-15 09:41  kps

	* gr.in, rialacha.meta.sed:
	  A bit of a hack to eliminate nested errors (=> valid XML error
	  markup, and also makes comheadan more robust).    Basically I
	  changed abairti to guarantee there is no extraneous markup
	  adjacent to a word (<line> was really the problem -- added a
	  space and a </line> tag at the end preceded by a space).     Then
	  in rialacha.meta.sed I only add error markup when the matched
	  phrase is not surrounded by markup (necessarily error markup).
	  The hack is because I need to move these non-markup characters
	  back outside of the new <E></E> tags immediately (so later lines
	  in rialacha.pl will see the markup!)

2003-07-14 22:36  kps

	* gin.meta.sed: deleted extraneous comment

2003-07-14 22:25  kps

	* README: Corrected web address.

2003-07-14 22:24  kps

	* Makefile: Changes for gin.meta.sed.

2003-07-14 22:20  kps

	* aonchiall.in, aonchiall.meta.sed:
	  Moved to using gin.meta.sed for uniformity, etc.  Also took the
	  opportunity to make the aonchiall grammar a bit more
	  sophisticated, b/c of some of the disambiguation rules I have in
	  mind.

2003-07-14 22:19  kps

	* rialacha.meta.sed:
	  Added gin.meta.sed, which basically amounts to what used to be in
	  rialacha.meta.sed.   All but last line now gone.

2003-07-14 21:46  kps

	* eisceacht.in, eisceacht.meta.sed:
	  Oops, I forgot to "cvs add" these a day or two ago when they were
	  added to the project.   Now they are in the new format which
	  relies on gin.meta.sed.

2003-07-14 21:44  kps

	* gin.meta.sed:
	  Combined common features of the three .in file grammars and put
	  the common meta-parsing in a single file for uniformity/ease of
	  change.

2003-07-14 16:33  kps

	* triail: Beginning of triail finished, through "chomh" rule.  All
	  tested and the ones not requiring subtle disambiguation all work.

2003-07-14 16:32  kps

	* rialacha.in:
	  Liberalizations here and there in order to make errors from
	  "triail" appear as they should.

2003-07-14 16:32  kps

	* rialacha.meta.sed:
	  Fixed bug in [^] syntax.

2003-07-14 16:31  kps

	* Makefile:
	  added "triail.err.old" and diff to "triail.err" target.

2003-07-14 12:10  kps

	* rialacha.meta.sed:
	  New macro for past autonomous verbs.

2003-07-14 12:10  kps

	* cuardach.c:
	  Restructuring of dictionary lookup to deal with Uppercase more
	  cleanly.  Was causing bugs with words like "San", which were
	  coming up with only the parts of speech of the capitalized word
	  in the db (a noun in this case) when in reality I should have
	  also been looking up the lowered version (and finding it's a
	  preposition in this case).

2003-07-14 12:08  kps

	* rialacha.in: Lots of bug fixes based on testing with OF81.
	  Mostly getting rid of false positives.   e.g. no mutation of past
	  autonomous in most cases.

2003-07-13 15:11  kps

	* cuardach.c:
	  Added "O" tag.  More information in the logs for "rialacha.*"

2003-07-13 15:10  kps

	* rialacha.in, rialacha.meta.sed:
	  Added the "ANYTHING" variable and eliminated the <X></X> syntax,
	  unifies the approach.

	  Changed UNMUTATED rule to ignore initial "M" since I think "ar an
	  margadh" etc. are acceptable.

	  Split the one-letter XML tags into the "ANPV" (allow attributes)
	  and the rest (which never allow attributes).	 Added new tag "O"
	  for pronomials since it belongs in the first class but "S"
	  (prep.) belongs in the 2nd.	Now rialacha.meta.sed generates a
	  simpler pattern for tags in the second class, including no
	  ambiguities necessary.  Gives a 30-40% speed improvement in
	  informal benchmarks.

2003-07-13 13:26  kps

	* gr.in:
	  One new message (PREFIXH) and additions for "eisceacht.*"

2003-07-13 13:25  kps

	* aonchiall.in:
	  Some additions, and corrected web page address.

2003-07-13 13:25  kps

	* cuidiu.html, iompar.html, sonrai.html:
	  General tidying of the exposition, added bits to sonrai for
	  "eisceacht.in" stuff.

2003-07-13 13:22  kps

	* Makefile:
	  Additions for "eisceacht.*", "iompar.html", and added cabhair.c
	  to the distributions for use by developers.

2003-07-13 13:21  kps

	* rialacha.meta.sed:
	  New abbreviations needed for latest rules added.

	  Also a minor bug fix -- in "unlenited" abbreviations, don't want
	  to trap words with initial "b'", "d'" or "m'"!

2003-07-13 13:20  kps

	* rialacha.in:
	  Lots of new rules.   First serious pass through all of CB94.

2003-07-06 22:41  kps

	* cuidiu.html, index.html, iompar.html, samptemplate.html,
	  sios.html, sonrai.html: added link for iompar.html to every page.

	  changed "Sponsor" link to "Projects" -> /gaeilge.html

2003-07-06 22:39  kps

	* iompar.html:
	  adding to web page, stuff about porting to other languages

2003-07-02 23:09  kps

	* COPYING, Makefile, README, aonchiall.in, aonchiall.meta.sed,
	  cabhair.c, cuardach.c, cuidiu.html, gr.in, hilite.awk,
	  index.html, rialacha.in, rialacha.meta.sed, samptemplate.html,
	  sios.html, sonrai.html, triail: Importing gramadoir sources

2003-07-02 23:09  kps

	* COPYING, Makefile, README, aonchiall.in, aonchiall.meta.sed,
	  cabhair.c, cuardach.c, cuidiu.html, gr.in, hilite.awk,
	  index.html, rialacha.in, rialacha.meta.sed, samptemplate.html,
	  sios.html, sonrai.html, triail: Initial revision

