2003-11-09 18:04  ed

	* README: Version 3.2.06.epa7.

2003-11-09 17:27  ed

	* findaffix.X, munchlist.X: Fix (based on Debian's) for insecure
	  temporary file created in findaffix and munchlist.

2003-08-24 12:50  ed

	* Makefile, Makekit, README, local.h.bsd, local.h.linux,
	  local.h.samp, local.h.sysv: Replaced local.h.samp with
	  local.h.{bsd,linux,sysv}.

2003-08-24 12:11  ed

	* test_inst: To avoid warnings during the test suite, add '#undef
	  foo' before each occurrence of '#define foo'.

2003-08-24 11:40  ed

	* README.SuSE: Updated description: the file gives changes from
	  epa1 to epa2.

2003-08-23 11:56  ed

	* Makefile, Makekit, buildhash.c, config.X, correct.c, ispell.1X,
	  ispell.5X, ispell.c, local.h.samp, lookup.c, parse.y, proto.h,
	  term.c, test_inst, languages/Makefile, languages/fix8bit.c,
	  languages/altamer/Makefile, languages/american/Makefile,
	  languages/british/Makefile, languages/dansk/Makefile,
	  languages/deutsch/Makefile, languages/english/BUGS,
	  languages/english/Makefile, languages/english/english.5X,
	  languages/english/msgs.h, languages/espanol/Makefile,
	  languages/francais/Makefile, languages/nederlands/Makefile,
	  languages/norsk/Makefile, languages/portugues/Makefile,
	  languages/svenska/Makefile, languages/test_dict/Makefile: Purging
	  my old email address.

2003-08-23 11:55  ed

	* mkdist: Adding 'mkdist' to create a distribution directory.

2003-08-23 11:55  ed

	* README: Preparing for version 3.2.06.epa6.

2003-01-18 15:43  ed

	* buildhash.c, languages/english/msgs.h: The macros
	  BHASH_C_NO_COUNT, BHASH_C_BAD_COUNT, and BHASH_C_ZERO_COUNT,
	  which are error message strings, now contain %s to give a
	  filename.  msgs.h for other languages will need to be updated
	  accordingly.

2003-01-18 15:41  ed

	* proto.h: The recently-introduced intptr_t typedef clashes with
	  _inttypes.h which seems to be included somehow on HP-UX.
	  Protected it by checking __INTTYPES_INCLUDED.  It's possible
	  other OSes will require similar changes.

2002-10-24 20:20  ed

	* parse.y: Patch from Tushar T to make compile with bison-1.75, by
	  removing a single stray semicolon.

2002-10-17 19:22  ed

	* README: Updated for version 3.2.06.epa5.

2002-10-17 15:34  ed

	* languages/english/: altamer.0, american.0, american.1,
	  american.2, british.0, british.1, british.2, english.0,
	  english.1, english.2, english.3: Fixes for more problems reported
	  by Richard Oudkerk.

	  >"american.2:improvizatorize/S" is wrong.  It should be
	  >"american.2:improvisatorize/S" (since english.2 contains
	  improvisator).

	  Yup.

	  >british.2 wrongly contains both "anaesthetisation/MS"
	  >and "anaesthetization/MS".

	  -ization removed.

	  >"anaesthetiser" is missing from british.* (although we have
	  >"american.0:anesthetizer/MS")

	  Well it's not missing, it is provided by the /R flag to
	  'anaesthetise'.  However the words coming from anaesthetiser with
	  the /M and /S flags are missing, so it needs to be a separate
	  entry.

	  >Since we have "american.2:fibrize/DGRSZ" we should probably add
	  >"american.2:fibrization".

	  For all words like this I've made sure there are the four
	  entries:

	  fibrize/GRSZ fibrized/U fibrizable fibrization/MS

	  >Since we have "american.2:fiberizer/MS" we should probably add
	  >"british.2:fibriser/MS" (and "american.2:fibrizer/MS").

	  I've added these flags for every -izer word (so removing the /R
	  flag from the -ize word and adding a separate -izer/MS entry if
	  necessary).  Noticed while doing this that 'enterpriser'
	  shouldn't be there.

	  >Should "british.2:gormandise/DGRS" be added? (english.2 contains
	  >"gormandise" with no flags).

	  british.2 already does have gourmandise/DGS and I made sure it
	  had gourmandiser/MS in the above step.  But you asked about the
	  kind with 'gorm' not 'gourm'.

	  I'll go and dig out the data files that come with respell.  They
	  say:

	  # Gourmand/gormand, a glutton: Ispell lists gourmand as english,
	  and # does not list gormand.	Collins lists gormand as a variant,
	  the OED # does not list it.  #  # Gourmandise: this is not in
	  fact gourmand + ize, but # a word ending in 'ise' like expertise.
	   As having expertise means # being an expert, so having
	  gourmandise means being a gourmand - so # gourmandise means love
	  of good food.  Confusingly, this is also # sometimes spelt
	  gormandize.  #  # Gormandize: this _is_ an -ize word, and the
	  pronunciation reflects # this.  It means to eat greedily.  #	#
	  Annoyingly, the OED has the two words immediately above both
	  spelt # as 'gormandize'.  (It also lists 'gorman' and 'gormande'
	  as # alternatives for gourmand.)

	  So what does this mean for british spelling?	I think it means I
	  should remove 'gourmandise' as a verb and replace it with
	  'gormandise'.  Well actually, I think I'll put gourmandise (the
	  noun) into english.

	  >Isn't "medieval" etc. acceptable in british.*?

	  Yes, but if ispell doesn't have it already I won't add it, since
	  differing variant spellings are a Bad Thing in a spelling
	  checker.

	  >Also we have different ordered flags in "misorganize/DGRSZ" and
	  >"misorganise/GDRSZ" :-)

	  I've put all flags into alphabetical order.

2002-10-17 12:55  ed

	* languages/english/: american.0, american.1, american.2,
	  british.0, british.1, british.2, english.0, english.1, english.2,
	  english.3: Fixes for more mistakes reported by Richard Oudkerk.

	  >"english.2:pediatrist", but no "paediatrist"
	  >"english.3:cyclopedia", but no "cyclopaedia"

	  I've sorted out all the -paed- words and split them into british
	  and american, with the exception of two which are always short-e
	  and always written 'ped': pederast and pedagogue (which needs
	  splitting anyway).  Except that keeping backwards compatibility
	  I've allowed -pedia in english and -paedia in british.

	  >"english.2:disklike", but no "disclike"

	  Added.

	  >"english.3:primaeval", but no "primeval"

	  All such words are now -eval english, -aeval british (so british
	  allows both for backward compatibility).  Except 'coeval' which
	  always seems to get E.

	  >"english.3:storeyed", "english.3:multistorey",
	  "english.3:blindstorey"
	  >		    but no "storey" (and "multistory" is american
	  only?)

	  I've moved all the 'storey' words into british, and 'story' when
	  it relates to buildings into american.  Exception: clearstory and
	  clerestory which apparently are both acceptable spellings but not
	  allowed with -storey.

	  >"british.1:arsehole/MS", but "english.3:arse"

	  I've put arse and ass on the same footing (in english.0).  I
	  think American users are unlikely to use the wrong spelling for
	  'ass' especially when you consider rhotic pronunciation.

	  >"american.0:aging/U" is acceptable in british also.	(Chambers
	  lists
	  >this before "ageing")

	  But ispell should allow only one of the variants.  Maybe I
	  wouldn't remove aging if it were already in there, but I'm not
	  going to add it.

	  >"extendable" and "extendability" are alternatives to
	  >"english.0:extendible/S", "english.1:extendibility".  (Chambers
	  lists
	  >the "a" spellings first.)

	  I don't think the difference in pronuncication is enough to add
	  these as separate entries.  Hmm, but ispell's current list of
	  words is a mess.  So I think I will have extendible, extensible
	  and extendable on equal terms for now.

	  >"english.2:medievalise/S" should be in british.2

	  Fixed above.

	  >Chambers only has "paralleled", "paralleling", but ispell has:
	  >"british.0:parallelled/U", "british.0:parallelling" instead.

	  Ick.	This word is the exception to the L-doubling conventions in
	  british spelling; four Ls so close together were thought to be
	  excessive I suppose.	So paralleled and paralleling can move into
	  english with the two-L variants disappearing.

	  >British spelling is "wilful", american "willful", but we have
	  >"english.0:willful/PY", "english.1:wilful/Y"

	  Made wilful british, willful american.

	  >British contains "skepticise" but not "scepticise".
	  "skeptic/sceptic"
	  >"skeptical/sceptical" are all in english.*, but have different
	  flags.
	  >(Is "skeptic" allowed in British?)

	  I think skeptic is used for someone who doesn't believe in UFOs
	  (after the American publication 'The Skeptical Inquirer').
	  Sceptic for all other senses.  This might not be a strong enough
	  argument to add skeptic if it didn't already exist, but I might
	  as well keep it.

	  So skeptic english, sceptic british.

	  >"english.2:estrogen" (and derivatives, "estrus") should be in
	  american?
	  >"english.3:oestrogen" should be promoted?

	  FWIW 'oestrus' is not derived from 'oestrogen', it comes from the
	  Latin for gadfly and is a much older word, according to the OED.
	  But it's not explained what the origin of 'oestr' is in
	  'oestrogen'.

	  I checked all words containing -estr- and after discarding a few
	  obvious ones which aren't related, got the following lists:

	  To split into OE british, E american: estrogen estrogenic
	  estrogenically oestrogen

	  Not related, leave alone: palaestra campestral estrade oestrus

	  Unknown (not in my 1970s OED or its addenda), leaving: estral
	  estray estrous estrum anestrus anoestrus diestrus dioestrus
	  estradiol estragon estrin estriol mestranol metestrus metoestrus
	  oestrin oestrone oestrous estrone estrus proestrus

	  >These should be sorted out:
	  >	    "american.2:esophagus", "british.2:oesophagus",
	  >	  "english.1:esophagi", "english.2:esophageal"

	  Moved into american.1 and british.1 for (oe|e)sophag(us|i),
	  american.2 and british.2 for (oe|e)sophageal.

	  >I don't think "english.2:topfull", "english.1:bellyfull" are
	  >allowable variants of "english.2:topful", "english.0:bellyful"

	  Agreed, it looks like a mistake.

	  >(These are the only ispell words apart from "full" which end in
	  >"-full", but there are dozens of words ending in "-ful", and I
	  can't
	  >find them on the dictionary.com)

	  Given how productive the -ful suffix is, it's unreasonable to
	  expect all of them to be listed.  None of the words you found
	  looks like a mistake, although I agree that some of them are a
	  bit exotic.

	  >Maybe have "instal" as alternative british variant of "install"

	  No, I'm not going to add variants unnecessarily.

	  >Maybe have "focussed/focussing" as alternative british variant
	  of
	  >"focused/..."

	  Strangely enough, this is the first thing I wrote to the ispell
	  mailing list about.  Again, it shouldn't be added because a
	  spelling checker should be consistent.

	  >Maybe promote "english.2:spelt/R" (and perhaps also
	  "english.3:misspelt")

	  To avoid confusion I have put spelt on the same level as spelled,
	  ditto misspelt and respelt.  (And noted that 'ispell' is listed
	  in english.0!)

2002-09-03 22:11  ed

	* README: Updated for 3.2.06.epa4 release.

2002-09-03 22:05  ed

	* test_inst: Removed dependency on DB_File: now we just read all of
	  local.h into a big array, modify the array, and write it out
	  again when done.  It turns out that this is all that the DB_RECNO
	  mode did anyway.

2002-09-03 19:42  ed

	* languages/english/: american.1, american.2, british.0, british.1,
	  british.2, english.0, english.1, english.2, english.3: Further
	  corrections prompted by Richard Ouderk.

	  Made wagon/waggon consistent (wagon english, waggon british).
	  Split faeces/feces into british and american.  Put whiz and whizz
	  both at level 0, pending eventual removal of one.  Dealt with
	  grey-/gray- words, noting that some like grayling and greyhound
	  do not have two versions.  (Following ispell's existing practice,
	  both grey and gray are allowed in english.)  Made hiccup english,
	  with hiccough kept in british for compatibility.  Sorted out
	  encyclopaedia by putting E in english and AE in british, in all
	  cases.  The incorrect spelling foetus has been relegated to
	  british, for compatibility only, with fetus moving into english.
	  Caddie and caddie are now on the same level.	Dealt with graecize
	  and graecism by allowing E or AE for english and always having
	  small G.

2002-09-03 17:22  ed

	* languages/english/: american.1, british.0, british.1, british.2,
	  english.0, english.1, english.2: Small fixes to the corrections I
	  made last time, following more discussion:

	  It's debatable whether there is any real american/british
	  distinction with revery/reverie, so I've put them both into
	  english.1.

	  'To kerb', meaning to place a kerb round a road or other open
	  space, has (re)appeared in british.

	  'Motorway' and 'windscreen' are common words, and not lookalikes
	  for commonly-misspelt words, so they've moved into levels 0 and 1
	  respectively.

	  Reinstated 'reprogramme' and 'preprogramme' as verbs in british.

	  Moved 'whilst' to level 0, since that has 'amongst', 'amidst'
	  already.

2002-09-03 00:33  ed

	* Makefile: Use set -e for those long shell commands, so that the
	  failure of one part gets noticed.

2002-09-03 00:29  ed

	* languages/english/: altamer.0, altamer.1, altamer.2, american.0,
	  american.1, american.2, british.0, british.1, british.2,
	  english.0, english.1, english.2, english.3: Richard Oudkerk
	  pointed out some errors in the wordlists, found by comparing
	  british and american.  To describe the changes made, it's best to
	  just include the two messages I sent him in reply:

	  Date: Tue, 3 Sep 2002 00:17:03 +0100 (BST) From: Ed Avis
	  <ed@membled.com> To: Richard Oudkerk <oudkerk@maths.man.ac.uk>
	  Subject: Re: inconsistencies in patched ispell dictionaries
	  Content-Type: TEXT/PLAIN; charset=US-ASCII

	  >I have noted what seem to be a few inconistencies in the english
	  >dicionaries you have in your 3.2.06.epa3 patches of ispell:

	  I've been through these and fixed most of them (for some value of
	  fixed, which is influenced by my goal of not doing anything too
	  radical).  I should make an epa4 release fairly soon.

	  >english.0:cozy/PRS		  should it be in american?

	  Done.

	  >english.1:cosy		  should it be in british? or
	  altamer?
	  >				  (americans seem to accept cosy as
	  a variant
	  >				  of cozy)

	  I'll take your word for it: I've added it to british and altamer.

	  >decolonised in british.2 only
	  >decolonize/DGS in american.2 only
	  >decolonise in english.2		 (should be in british?)

	  Done.

	  >agonisingly in british.0 only
	  >agonizedly in american.0 only
	  >	     (agonizedly does not exist according to
	  dictionary.com,
	  >	     and google.com gives 271 matches for agonizedly versus
	  >	     29200 for agonizingly---if agonizedly is acceptable it
	  is
	  >	     probably too rare for american.0)

	  Fixed, but I kept agonizedly because it seems plausible, and
	  there's no harm in leaving it in american.0 since it's not close
	  to any common misspelling.  (A word 'teh', on the other hand,
	  would have to be kept out of american.0... you see what I mean.)

	  >realisables/U in british.0 only
	  >		  (Most/all of the 368 matches for this on
	  google.com are
	  >		  form French web pages.  Most/all of the 7880
	  matches for
	  >		  realizables are from Spanish webpages)

	  But I will leave it in because if you did use the word in
	  English, you'd want to spell it correctly :-).  Ispell is a
	  spelling checker not a usage checker, so it's okay to have
	  foreign words if they might sometimes be used as English words.

	  >organised/AU in british.0 only
	  >unorganized/PY in american.0 only
	  >		 (check all words derived from organize/organise)

	  Fixed.

	  >metallisation/MS in british.0, but metallization/MS in
	  american.1
	  >(metallise/DGS and metalize/DGS in british.2 and american.2)

	  Standardized metallize and related words.

	  >packetisation in british.1, but packetization in american.2

	  Moved it to american.1 (preferable to downgrading packetisation
	  to british.2, which would break compatibility).

	  >isomerizeparabolization    is it a word?

	  How should I know? :-)

	  >only 5 matches on google, and each of these appears to be from a
	  >wordlist file,

	  Heh :-).

	  >I would suggest just removing it.

	  TODO

	  >misprise/DGS in british.2 only (but obscure)

	  Hmm.	Currently ispell has misprise british, misprize english.
	  Collins says: misprize or misprise.  Chambers prefers misprise.
	  And the OED allows only the Z spelling, since the it comes from
	  the same French verb as 'prize', and we write that with a Z.	But
	  misprision is always with S.

	  I would incline to allow only the Z spelling, but since at least
	  one British dictionary prefers S and ispell already allows it for
	  british, I'll keep it there.	So Z is always allowed, and S if
	  british.  In other words keep the status quo!

	  >renationalize/G in british.2 only

	  british.2 has renationalise/G, but that's redundant since
	  nationalise in british.0 has the A and G flags.  So I've removed
	  it.

	  >The following words have different flags in american.2 /
	  british.2
	  >
	  >antithesize/S	  antithesise
	  >disequalize/MRSZ	  disequalise/RSZ
	  >gourmandize/DGS	  gourmandise/DGRS
	  >Moravianized/MS	  Moravianised/S
	  >vaporizable/MS	  vaporisable/S
	  >volatilize/DGRSZ	  volatilise/GRSZ

	  Fixed by adding flags where needed.  Except for the M flag which
	  I've sometimes removed (disequalize's makes no sense, but
	  vaporisable's I have allowed because of existing precedent).

	  -- Ed Avis <ed@membled.com>

	  Date: Tue, 3 Sep 2002 00:18:16 +0100 (BST) From: Ed Avis
	  <ed@membled.com> To: Richard Oudkerk <oudkerk@maths.man.ac.uk>
	  cc: ed@membled.com Subject: Re: inconsistencies in patched ispell
	  dictionaries Content-Type: TEXT/PLAIN; charset=US-ASCII

	  >	  http://www.maths.man.ac.uk/~oudkerk/britishize.html

	  I'll have a look at this.

	  You sent some more corrections which I've dealt with:

	  >If I use a medium dictionary, I will sometimes only be offered
	  the
	  >american variants of certain words.
	  >
	  >Cases where the normal (or only) british variant is in
	  english.[2-3]
	  >(or absent), but normal american variant is in english.[0-1].
	  (The
	  >number before the colon indicates the dictionary file number.)
	  >
	  >British				  American
	  >----------------------------------------------------------------------
	  >2:ageing/U			  0:aging (age/DGRGZ)

	  Fixed by removing G flag and putting age?ing separately in
	  british.0 and american.0.

	  >3:chilli			  1:chili (+ 2:chilies)
	  >2:skilful/Y			  0:skillful/PUY
	  >3:unskilful			  0:unskillful (skillfull/PUY)
	  >2:smoulder			  0:smolder/DGS (0:smoldering/Y)
	  >2:instalment/S		  0:installation/MS
	  >2:artefact/S 		  0:artifact/MS

	  Fixed these by a straightforward american/british split (and
	  sometimes promoting to a lower-numbered file).  But I think you
	  mean 'installment' not 'installation' :-).

	  >1:furor/MS			  2:furore/MS

	  I'd consider these to be separate words because they are
	  pronounced differently (according to Collins).  So furore is a
	  Britishism and furor an Americanism.	Since ispell is a spelling
	  checker not a usage checker (for example, I moved 'airplane' and
	  'aeroplane' into english) it should allow both variants.  But I
	  have put them both at level 1.

	  >2:instil/S			  1:instill/DGR

	  One L british, two Ls american (and the same for instilment).

	  >3:dependant/S		  0:dependent/ISY

	  -ant noun, -ent adjective.  Although the /S flag for dependent
	  suggests a use as a noun, perhaps in US.  And there is 'an
	  independent' of course.  That is my own feeling, but is it right?

	  The OED says dependant or -ent for the substantive, saying, 'The
	  spelling -ent, after Latin, is less usual in the substantive;
	  confer "defendant", etc.'.  The adjective is the other way
	  around, with -ent more usual.

	  Collins gives -ant for the noun, with -ent a variant spelling
	  'esp. US'.  For the adjective it says -ent, but 'US (sometimes)
	  -ant'.  So it looks like the Americans just have a general
	  confusion about which is which.

	  Chambers agrees with the OED.

	  So ispell is correct to have both spellings in english - but I
	  have moved dependant from 3 to 0 to avoid confusion.	(Imagine
	  typing, correctly, 'dependant' and having ispell tell you it
	  should be 'dependent'.)

	  Also I have removed 'dependents', hoping that American users
	  won't mind the -ent/-ant distinction being inflicted on them, but
	  I've added 'independents' as a special case.

	  >2:liquorice			  1:licorice

	  Split into two.  BTW Collins says licorice is usual in Canada.

	  >2:pedlar			  0:peddler/MS

	  Hmm.	Collins says 'pedlar or esp. US peddler, pedler'.  Chambers
	  decides to distinguish between pedlar and the verb peddle, which
	  is a back-formation from pedlar and gives peddler.  The OED says
	  pedlar, also US pedd?ler.

	  I think that since Chambers thinks peddler is a different word,
	  two steps away from pedlar, and since it may be pronounced
	  differently (like fiddler, not like butler), both should be in
	  english.  But for fairness I'll put them both on the same level
	  so people with small dictionaries don't get told that pedlar is
	  wrong and peddler correct.  I will not introduce the spelling
	  pedler.

	  >2:polythene			  1:polyethylene

	  I think this is like teflon versus tetrafluorethane, or vinyl
	  versus polyvinylacetate or whatever it is.  They're pronounced
	  differently and have slightly different uses (everyday vs formal)
	  so keep both.

	  >2:revery			  1:reverie/S

	  Well the OED allows only reverie, while Collins and Chambers
	  allow both but prefer reverie.  I'll take your word for it that
	  the american spelling is revery, and move these into american.1
	  and british.1.

	  (If it turns out that revery is just a variant and not
	  particularly american, then it could be removed to enforce
	  consistency.	Although I might not want to do that myself.)

	  >2:phial			  0:vial/MS

	  The OED has these two as separate, with different pronunciations,
	  although sharing the same root and having similar meanings.
	  ('Phial' is etymologically the better spelling.)  Collins says
	  that vial is a less common variant of phial, again with its own
	  separate entry and different pronunciation.  Only Chambers says
	  flatly that vial is 'same as phial', but even there the
	  pronunciation is different.

	  Since the two British dictionaries allow both forms, we can say
	  that ispell should allow both for british.  Should it disallow
	  one of them for american?  I can't see any hints that it should,
	  although I do not have a US dictionary.

	  >3:kerb/G			  0:curb/DGS

	  Moved kerb to british.0 (and deleted 'kerbing' which seems
	  anomalous); curb/DGMS stays in english.0.

	  >3:bobsleigh			  1:bobsled (1:bobsledding
	  2:bobsledder)

	  This should follow the same pattern as sleigh/sled, for which
	  ispell makes no british/american distinction.  Rightly, IMO,
	  since if you want to write one of these words then ispell's job
	  is to check its spelling, not berate you for using the wrong
	  word.  The two are not variant spellings of the same word; they
	  are different, like aeroplane and airplane.

	  However I have taken the opportunity to sort out the
	  sled/sleigh/sledge words and add missing flags.  They're all
	  regular now except that there is no word 'bobsledge'.

	  >3:haemorrhage		  1:hemorrhage/DGS

	  Hmm, these words don't appear to have been split properly between
	  british and american.  I'll do the haem- words but it would take
	  forever to do all words with AE or OE; I have done that for my
	  respell project and I would probably borrow that list instead, to
	  do the job properly.

	  Noticed ispell had the erroneous spellingh haemacytometer for
	  haemOcytometer.  I had to guess some of these since not all the
	  medical terms are in the dictionary, and some might be like
	  hemisphere.  Removed 'haem' which is not in the dictionary.
	  Fixed haemachrome (wrong for haematochrome, according to the
	  OED), haemacytometer (should be haemo-), fixed initial capital
	  letter in haematoxylon.

	  >2:titbit			  1:tidbit/S

	  Hmm, not sure about this.  These are different pronunciations,
	  but only just.  I think I will leave both available but on an
	  equal footing (level 1).

	  >2:motorway			  0:freeway/MS
	  >2:windscreen 		  1:windshield

	  This is not a question of spelling.

	  >2:carat			  1:karat

	  Collins says that these are different.  A carat is either a
	  measure of the weight of a diamond (always spelled 'carat'), or a
	  measure of the proportion of gold in an alloy ('carat' UK,
	  'karat' US).	Chambers agrees.

	  So ispell should have carat english, karat american.	I don't
	  think there is much point having the two meanings at different
	  dictionary sizes, that would just confuse.  So carat goes in
	  english.1 and karat in american.1.

	  >(3:kerbstone 		  2:curbstone)

	  Since this refers only to kerbs at the edge of a road, and is not
	  related to other meanings of the verb 'curb', it splits cleanly
	  into american and british.

	  >(3:Caesarean 		  2:Caesarian/M)

	  But there is also the AE / E distinction to worry about.  Ispell
	  currently has all four combinations in english!  And then should
	  it be small C or capital C?  Ispell chooses Caesar- but cesar-,
	  which makes some kind of sense.

	  I won't do anything to these apart from put them all in the same
	  place.  Maybe later there can be a definitive ruling on which is
	  correct.

	  >(3:haemophilia		  2:hemophilia)
	  >(2:hemophiliac		  3:haemophiliac and lots more
	  hemo-words)

	  Fixed above.

	  >Variants in english.* which should perhaps be unacceptable in
	  british
	  >(or made "rarer" than the british spelling)
	  >-----------------------------------------------------------------------
	  >2:caldron/S
	  >0:mold/DGRSZ (0:molder/DGS, 0:molding/A 0:moldy/PR,
	  >					  1:moldboard 2:moldable)
	  >2:molt/RS
	  >1:vise/DGV (1:viselike)
	  >0:willful/PY
	  >0:prolog/MS
	  >0:dialog/MS

	  Agreed, all these should be in american only.  Except for caldron
	  which I'm really not sure about; all three of my dictionaries
	  accept it as an alternative to cauldron without saying it is US.
	  So if it needs to be in the ispell wordlist - which it does,
	  since I don't have the guts to remove it - then it should be on
	  the same level as cauldron.  Maybe later one of the two can be
	  removed.

	  Mold / mould I have split into american and british.	I have a
	  feeling mould should be in altamer but I don't have any proof.
	  I've removed 'detmold' which is perhaps a lowercased version of
	  Detmold, a city in NW Germany (Collins).

	  >jewellery is in english.2, but jewelry is in american.0

	  I've moved jewellery and jewelery into level 0.

	  >carburettor is not included but we have english.2:carburetor/S,
	  >					  english.3:carburetter

	  Hmm.	Collins says carburettor, carburetter, or US carburator.
	  The OED says -ettor or -etter.  Chambers says -ettor, -etter, or
	  US -etor, -eter.

	  I'll put carburettor in british and move carburettor to british
	  too.	Carburetor can move to american.  There is no need to add
	  the fourth spelling with -eter.  I've also dealt with
	  carburetting, carburetted, which have single T in american.

	  >cissy is not included, but we have english.2:sissy/S

	  The OED says cissy is a variant of sissy: same pronunciation,
	  same meaning.  Chambers favours the C-spelling, but agrees that
	  they are just variant spellings of the same word.  Collins agrees
	  with the OED.

	  So I think ispell's current policy is right.

	  >connexions is in english.3, but connexion is in british.2

	  Moved both to british.2.

	  >distill is in english.0, but distil is no where.

	  Split between american and british; same for bidistil.

	  >archeology is only in altamer.0, but we have
	  english.0:archaeology

	  I think this is deliberate, the US spelling being normally with
	  AE.

	  >fledgeling is not in british.*, but we have
	  english.0:fledgling/MS

	  This should follow the same pattern as judgement.  Currently
	  ispell seems to have judgment english, judgement british and
	  altamer.  Not ideal - but a precedent for adding fledgeling to
	  british.0.

	  >paedophile is not in british.*, but we have english.2:pedophile
	  and
	  >					  english.3:paedophilia

	  I've split paed- words into british and american, except for
	  pedagogue and pederast, which for some reason (pronunciation?)
	  are almost always with E.  (Removed paederast from english.3.)

	  >reprogram is in american.0, but probably deserves to be in
	  >					  english.* as well
	  >
	  >preprogram is in american.2, but probably desrves to be in
	  >					  english.* as well

	  Agreed.  I've also got rid of reprogramme and preprogramme.

	  >whilst is in english.2, but is much more common in British than
	  >American

	  That's a question of usage not spelling.

	  >Some websites I looked at when checking more closely about
	  Canadian
	  >spelling:
	  >------------------------------------------------------------------------
	  >http://www.luther.bc.ca/~dave7cnv/cdnspelling
	  >http://www3.telus.net/linguisticsissues/BritishCanadianAmerican.htm
	  >http://www.peak.org/~jeremy/dictionary/spellcat.html
	  >http://ace.acadiau.ca/polisci/aa/digagora/tutorial/canas.html

	  Thanks, will look at these if I ever get round to making a
	  canadian wordlist.

	  -- Ed Avis <ed@membled.com>

2002-09-03 00:08  ed

	* languages/english/check_dups: It's confusing that there are
	  english.[0123] for the wordlist, but english.5 is a manual page!
	  At least it used to confuse check_dups.  Now there's a special
	  kludge to stop looking when it finds a 'gap', in other words
	  since english.4 is not there, english.5 cannot be part of the
	  wordlist.

	  Also fixed the comment at the top to match the program's name
	  :-).

2002-09-02 23:45  ed

	* languages/english/Makefile: Building the *.words files requires a
	  working ispell, and for that you need a dictionary hash file.
	  Rather than add a dependency across directories we just prompt
	  the user to symlink english.hash into the current directory.
	  (Good enough since this is a developers' facility anyway.)

	  Also, building the files will now stop at the first error instead
	  of continuing regardless.

2002-09-01 13:21  ed

	* README: Updated my website and email address.

2002-06-15 13:02  ed

	* README: Fixed small typos in release numbers.

2002-06-15 12:59  ed

	* README: Updated README for 3.2.06epa3.

2002-06-15 12:45  ed

	* languages/english/BUGS: This BUGS file is the remains of the todo
	  list - just one item is left.

2002-06-15 12:39  ed

	* languages/english/Makefile: Added a possible test for _all_
	  duplicates to the Makefile, but it's commented out at the moment.
	  This meant making the *.words files have duplicate lines when
	  the same word can be generated two ways.

2002-06-15 12:29  ed

	* languages/english/check_dups: Added a very basic usage message.

2002-06-15 12:27  ed

	* languages/english/: Makefile, altamer.2, american.0, american.1,
	  american.2, british.0, british.1, british.2, check_dups,
	  english.1, english.2, english.3: To check for 'obvious
	  duplicates' (same base word appearing in two files) wrote a Perl
	  script check_dups and a new 'check' target which runs it.

	  Using this tool, removed duplicates from some of the lists, and
	  words which were common to american and british have moved into
	  english.  The generated dictionaries are the same, apart from
	  'whiskies' no longer being allowed in american (since 'whisky' is
	  not).

2002-06-15 11:23  ed

	* languages/english/: altamer.0, altamer.1: Both analogue and
	  travelogue have been added to altamer, since they are commonly
	  used in US publications.

2002-06-15 11:21  ed

	* languages/english/: altamer.1, altamer.2: Since altamer has
	  draught, I've added the related words like updraught and
	  indraught, but only for the sense of 'windy' and not for words
	  like draughtsman.  I assume that draught in altamer is meant to
	  be used for the physical sense and draft for other senses, as in
	  British usage.  Could be wrong about this.

2002-06-15 11:17  ed

	* languages/english/: american.2, british.2, english.2: The K
	  spelling for words like decagram (dekagram) and decametre will
	  not be allowed, for consistency.  This makes sense when you
	  consider that for example decalogue has only the C spelling.

2002-06-15 11:13  ed

	* languages/english/: american.2, british.1, british.2, english.2,
	  english.3: Sorted out 'colorate' and derived words.  The correct
	  spelling is without the U in both british and american - so these
	  words have moved to english.	The -our spelling is sometimes used
	  as a variant, but for consistency ispell won't allow it.

2002-06-05 22:22  ed

	* languages/english/: altamer.1, altamer.2, american.0, american.2,
	  british.0, british.2, english.0, english.2, english.3: Fixing
	  some small errors found when diffing the new wordlists against
	  the old versions.

2002-06-05 22:18  ed

	* languages/english/Makefile: When generating *.words, put one word
	  per line and remove duplicates.  These files are now removed by
	  'make clean'.

2002-06-05 21:05  ed

	* languages/english/Makefile: New 'words' target builds lists of
	  all words in the various dictionaries.  I plan to use this for
	  testing my changes, looking at which words have been added to and
	  removed from each dictionary.

2002-06-05 20:20  ed

	* ispell.1X: Fixed typo and punctuation in description of -e5.

2002-06-05 18:18  ed

	* languages/english/english.3: Removed wrong 'concencus' - a
	  consensus is to do with consent.

	  But the same mistake was more deeply ingrained because 'concent'
	  was in there too!

2002-06-05 18:16  ed

	* languages/english/altamer.2: Since altamer has jewelled and
	  jewelling it should have bejewelled and bejewelling.	Similarly
	  jeweller gives jewellery.

2002-06-05 18:14  ed

	* languages/english/: altamer.1, altamer.2, american.2, british.2,
	  english.2: Words coming from libel have -ll- in british and
	  altamer, -l- in american.  This includes 'libellist' - I hope
	  that's correct.

2002-06-05 18:08  ed

	* languages/english/: american.2, british.2, english.2: Split
	  'homolog' and 'theolog' into british/american versions.  Although
	  I thought these could be -logue even in american?

2002-06-05 18:07  ed

	* languages/english/: american.2, british.2, english.2, english.3:
	  Split some -gog forms wrongly in english into -gogue and -gog.
	  But Memphremagog just needs a capital letter...

2002-06-05 18:03  ed

	* languages/english/: american.2, english.2: Moved analogism from
	  american to english.

2002-06-05 18:02  ed

	* languages/english/: american.2, british.2: Adding prologise to
	  british; different from prologuise because it is pronounced with
	  a soft G.

	  Committing misanthropize fix which got lost before.

2002-06-05 18:01  ed

	* languages/english/american.2: Whoops, really fixed misanthropize
	  this time.

2002-06-05 18:00  ed

	* languages/english/american.2: Fixed 'mizanthropize'.

2002-06-05 18:00  ed

	* languages/english/american.2: Removed 'labour's' from american -
	  probably added as a result of careless spellchecking on
	  Shakespeare :-P.

2002-06-05 17:59  ed

	* languages/english/american.2: Removed 'epoches' from american, we
	  have 'epochs' in english.

2002-06-05 17:58  ed

	* languages/english/american.2: Removing 'Chablises', it was only
	  in american and I'm not sure it's right.

2002-06-05 17:57  ed

	* languages/english/: american.2, british.2, english.3: Purged a
	  few stray 'centre' words from english and split them into
	  centre/center.

2002-06-05 17:55  ed

	* languages/english/british.0: Added missing 'burglarise' to
	  british.

2002-06-05 17:54  ed

	* languages/english/american.0: Removed /Y flag from agonizedly.

2002-06-05 17:54  ed

	* languages/english/: altamer.0, altamer.2, american.0, british.0,
	  english.0: The word is 'marshal' with marshalled, marshalling,
	  marshaller in british and altamer, single-L versions in american.

	  But also altamer has 'marshall'.

2002-06-05 17:50  ed

	* languages/english/: altamer.0, altamer.1, altamer.2: Both british
	  and american have crystalli[zs]e; altamer has crystalize.  And
	  english has crystalline while altamer has crystaline.

2002-06-05 17:42  ed

	* languages/english/: altamer.2, american.2, british.2, english.1,
	  english.2, english.3: 'glamour', 'glamorous' are normal
	  everywhere; 'glamor' is allowed by altamer.  And Glamorgan has a
	  capital G :-).

2002-06-05 17:39  ed

	* languages/english/: altamer.0, altamer.2, american.2, british.2,
	  english.2: Assuming that 'councillor' is the normal spelling
	  everywhere, with 'councilor' an american variant.  The old
	  wordlists seemed to indicate this although they weren't
	  consistent.

2002-06-05 17:37  ed

	* languages/english/: altamer.0, american.0, british.0, english.0:
	  Split panelist into panellist british/altamer, panelist american.

2002-06-05 17:35  ed

	* languages/english/altamer.2: Added missing 'empanelled' and
	  'impanelled' to altamer.

2002-06-05 17:34  ed

	* languages/english/: altamer.0, altamer.1, altamer.2, american.0,
	  american.2, british.0, british.2: Added some flags to altamer
	  which were present in the american equivalents.  Also sorted out
	  dial/dialling/redialler etc.

2002-06-05 17:21  ed

	* languages/english/: altamer.2, british.2, english.2: Sorted out
	  crenel / crenelle.  Before, both spellings were in english and
	  I've kept that because I don't know enough to say what the one
	  british and one american spelling should be.

	  With related words, we have

	      british: crenel -> crenelled, crenelle -> crenelled
	      american: crenel -> creneled, crenelle -> crenelled

	  So given that -elle is in english, the -ll- words should be too.
	  But american can derive -l- words from crenel, so those need to
	  be added to the american list.

2002-06-05 17:13  ed

	* languages/english/altamer.2: Adding 'houselled' to altamer since
	  'unhouselled' is already in there.  But I'm not adding
	  'houselling', because the spelling with one L better reflects the
	  pronunciation (HOUZE-ling).  It is LL in british, but that's too
	  bad.

2002-06-05 17:11  ed

	* languages/english/american.2: Removed mistake 'psycoanalyze'.

2002-06-05 17:11  ed

	* languages/english/english.1: Added psychoanalyses, plural of
	  psychoanalysis, to english.

2002-06-05 17:09  ed

	* languages/english/: altamer.1, english.0: Standardized
	  'nonprogrammer' to follow 'programmer'.

2002-06-05 17:08  ed

	* languages/english/british.2: Added missing 'practicalise' and
	  -isation to british.

2002-06-05 17:06  ed

	* languages/english/: american.0, american.1, american.2,
	  british.0, british.1, british.2, english.0, english.1, english.2,
	  english.3: Split all occurrences of sulphur into sulfur american,
	  sulphur british.

	  (I've heard that some chemists use the spelling with an F - but
	  while I would readily defer to their wisdom on matters of
	  chemistry, they are not necessarily experts on spelling.
	  'sulfur' is completely against British newspapers, Collins and
	  even the OED, so it shouldn't be allowed in british.	After all,
	  realize is not allowed in Ispell's british dictionary, neither is
	  fetus, and these are both considered acceptable British spellings
	  while sulfur is not.)

2002-06-05 16:34  ed

	* languages/english/: american.0, american.1, british.0, british.1:
	  Made consistent capitalization of Balkanize.

2002-06-05 16:32  ed

	* languages/english/: american.0, british.0: Removed 'offencive'
	  from british, 'defensed' and 'defensing' from american.

2002-06-05 16:30  ed

	* languages/english/american.2: Added missing parabolization to
	  american.

2002-06-05 16:28  ed

	* languages/english/: american.0, american.1, american.2,
	  british.2, english.0, english.1, english.2: 'Gram' is the more
	  common spelling everywhere; british additionally allows 'gramme'.
	  Made sure all the derived words have /MS flags.

2002-06-05 16:13  ed

	* languages/english/: american.2, british.0, british.2: Metre is a
	  noun, never a verb - so it shouldn't get the /DGMS flags.  Made
	  all the metre-derived nouns have /MS flags.

2002-06-05 16:06  ed

	* languages/english/: american.2, british.1, british.2, english.1:
	  Fixed some words which should be color- even in british spelling.
	   They were:

	  colorimetry colorimeter colorimetric colorimetrical
	  colorimetrically colorant decolorant colorate decolorate

	  I haven't moved them from british/american to english, maybe I
	  should but this way it's easier to make another policy change
	  later :-P.  discoloration however is now in english along with
	  coloration.

2002-06-05 16:02  ed

	* languages/english/american.2: 'decolorise' in american was surely
	  a mistake.

2002-06-05 15:58  ed

	* languages/english/: altamer.0, altamer.2, american.0, american.2,
	  british.0, british.2, english.0, english.3: Woollen and woolly
	  have two Ls in british and altamer, one L in american.

2002-06-05 15:54  ed

	* languages/english/: altamer.1, british.1: Removed 'tyne' since it
	  is an obsolete word and not worth checking.

2002-06-05 15:52  ed

	* languages/english/: american.2, british.2, english.0, english.1,
	  english.2, english.3: Standardized aeon/eon to have eon versions
	  in english, with british additionally allowing aeon.	Also some
	  sorting inconsistencies (sigh...)

2002-06-04 22:48  ed

	* languages/english/: altamer.0, altamer.1, altamer.2, american.0,
	  american.1, american.2, british.0, british.1, british.2,
	  english.0, english.1, english.2, english.3: Removed /D and /G
	  flags from vectori[zs]er.

	  Etna and Aetna: there's the volcano, which is spelled Etna in
	  both Britain and America (since the pronunciation is with a short
	  E sound).  Then there's an etna, which is 'a vessel used for
	  heating liquids' according to the OED (it's not in Collins), with
	  variant spelling aetna.  So Ispell's english dictionary should
	  have Etna and etna, with Aetna and aetna included in british.

	  'Tittuppy' is in the british list, but the OED says 'tittupy'
	  (not in Collins).  So tittupy moves from american to english and
	  tituppy disappears.  Likewise tittupping, tittupped, gossipping,
	  gossipped.  In fact only kidnapped/ing (which is a compound word
	  like inputting) and worshipped/ing (an odd case) get the double P
	  for an unstressed syllable in British usage.	(The double-P forms
	  valid in british are also included in altamer.)

	  Standardized plough british, plow american.  Might add plough to
	  altamer if I find more evidence.

	  Removed pointless /S from agonisingly.

	  Judgement and acknowlegdement are often written without an E in
	  modern British usage (and Collins agrees).  So judgment and
	  acknowledgment have moved from american into english.  (This
	  leads to the inconsistency of allowing both judgement and
	  judgment in british - but that's probably okay since this is the
	  best-known word with two variants in british.)  I applied the
	  same process to acknowledgement and so on, so the final version
	  is: -dgment in english, -dgement in british and altamer, for all
	  such words.

	  Nondefense and predefense are wrongly listed as english; they
	  should be american with -defence in british.

	  Removed some redundant entries for patroni[zs]ing.

	  Made flags for reconnoitre/reconnoiter consistent, also added
	  reconnoitrer to british.

	  Gourmandise: this is not in fact gourmand + ize, but a word
	  ending in 'ise' like expertise.  As having expertise means being
	  an expert, so having gourmandise means being a gourmand - so
	  gourmandise means love of good food.	This word has moved from
	  british to english.

	  Picked 'Romania' as the correct spelling for english.

	  Removed spurious /S from gruelling.

	  Made sure 'armour' is always -our in british and -or in american:
	  so these words can never appear in english.  However, 'armorial'
	  meaning 'heraldic' comes from a different root (with a different
	  pronunciation), so it and related words are in english with -or.

2002-06-04 22:04  ed

	* languages/english/: american.0, british.0: pyjamas are already
	  plural, so they don't get /S.  And 'pyjamaed' is not a word, or
	  if it is, the user can add it by hand.

2002-06-04 22:00  ed

	* languages/english/american.2: Removed 'dizequalise' and similar
	  words from american, they were surely caused by carelessly
	  regexping words to americanize the spelling.

2002-06-04 21:49  ed

	* languages/english/british.2: Following Collins, british spelling
	  has 'vaporise' not 'vapourise'.  But note this does not apply to
	  'colourise'.

2002-06-04 21:35  ed

	* languages/english/: american.2, british.2, english.0, english.2,
	  english.3: 'seisin' and the legal sense of 'seise' are now
	  allowed both S and Z in both british and american.  Which is not
	  ideal, but I didn't feel qualified to make any stricter
	  judgement.

2002-06-04 20:16  ed

	* languages/english/: american.0, british.0, english.0: The
	  mainstream spelling is 'wagon' everywhere.  The british form
	  'waggon' will not be supported by ispell.  (Picking just one in
	  order to enforce consistency.)

2002-06-04 20:10  ed

	* languages/english/: altamer.0, altamer.2, american.0, american.2,
	  british.0, british.2, english.0, english.1, english.2, english.3:
	  Fixed some inconsistencies with tranquillize, tranquillity etc.
	  by picking two Ls for british and one or two for american.
	  Following the established pattern I've made the two-L variant
	  altamer and one-L the mainstream american, but I don't know
	  whether this is correct.

	  Also a lot of spurious changes from slightly different sort(1)
	  programs.  How I loathe locale-based sorting...

2002-06-04 18:18  ed

	* languages/english/: american.0, english.0: Moved 'jewelry' from
	  english to american, because it is not used in british spelling,
	  not even when quoting a speaker using that pronunciation.

2002-06-04 18:15  ed

	* languages/english/: american.0, american.1, american.2,
	  british.0, british.2, english.0, english.1, english.2: There were
	  some words listed in american which weren't in english because
	  they were seemingly Americanisms.  But because ispell's job is to
	  check spelling, not vocabulary, I've moved them into english.
	  Similarly some british-only words have moved into english.

	  The words moving from american are:

	  airfoil airplane bucketfuls catheti hydrocaryaceous illutation
	  Italianism mazurka Melampus metagram signatories reprogram
	  rencounter

	  and those from british:

	  amorist practician misemphases tining aerostation aerodrome
	  crystallographical (a single-L version is already in altamer)
	  aerophagia aeroplane aerometry aerogel aeropropulsion aerophyte
	  aerodonetics hypercatharses furore orientate initially
	  aeromechanic aerofoil aerophobia

	  There are also some words where the american and british
	  spellings reflect a slightly different pronunciation.  For
	  example mustache versus moustache.  Personally I would write
	  mustache when quoting an American speaker, but I have a feeling
	  that most British newspapers would change the word to moustache,
	  even though its pronunciation is different.  If you agree with
	  that policy, then it makes sense to keep words like specialty out
	  of british and speciality out of american.  For the record, the
	  words I found were:

	  moustache speciality subspeciality learnt parametrize (as opposed
	  to parameterize)

	  mustache specialty subspecialty

	  But 'asshole' I did move because it's a common spelling in both
	  systems.

2002-06-04 18:06  ed

	* languages/english/: american.0, american.2, british.0, english.0,
	  english.2: Moved siphon from american to english, since it is the
	  most common spelling in Britain too.

	  Following Collins, the british plural of whisky is whiskies.

2002-06-04 18:02  ed

	* languages/english/british.0: Following Collins, changing
	  'whiskys' to 'whiskies'.

2002-06-04 18:01  ed

	* languages/english/: american.2, british.2: Removed /M from
	  unsceptre since it is not a noun.  Added /D flag to british.

2002-06-04 18:00  ed

	* languages/english/: altamer.0, altamer.2, american.0, american.2,
	  british.0, british.2, english.3: Made some more flags consistent;
	  moved 'metacentres' out of english and split it into american and
	  british variants.

2002-06-04 17:47  ed

	* languages/english/: british.0, british.1, british.2: Added /M
	  flag to lots of -able words which had it in american but not
	  british.  Also removed redundant 'reacclimiatisable' from
	  british: acclimatisable has the /A flag already.

2002-06-04 17:25  ed

	* languages/english/: american.2, british.0, british.2: Manually
	  fixed the flags for a couple of -able words.	Will try to use
	  reconcile_flags (a Perl script I wrote) for the others.

2002-06-04 17:20  ed

	* languages/english/: american.0, british.0: Removed /U flag from
	  reali[zs]es, since 'unrealize' is not a verb.  (Or if it were, it
	  should be more generally available.)

2002-06-04 17:17  ed

	* languages/english/: british.0, british.2: Removed some duplicates
	  which I accidentally introduced.

2002-06-04 17:12  ed

	* languages/english/: american.0, american.1, american.2,
	  british.0, british.1, british.2, english.0, english.1, english.2,
	  english.3: Fixed british 'licenseable' to 'licensable' following
	  Collins.

	  Moved 'licenseless' from english to american.

	  Split lots of -ize words which were wrongly in english into -ize
	  american and -ise british.  Also 'recognizee'.

	  Some small sorting changes, 'sort --ignore-case -k1,1 -t/' seems
	  to reproduce the original order but it's not quite right.

2002-05-19 18:47  ed

	* README.SuSE: Fix minor spelling mistake.

2002-05-19 18:46  ed

	* README, README.SuSE: Updated README for 3.2.06epa2 release; new
	  (and temporary) file README.SuSE explains what I decided for each
	  patch from SuSE.

2002-05-19 18:42  ed

	* buildhash.c, config.X, lookup.c, proto.h: Most of SuSE's
	  ispell-3.2.06-types.patch: the macro intptr_t defines a type
	  large enough to hold a pointer, and the standard limits PATH_MAX
	  and NAME_MAX are used if available.

2002-05-19 18:26  ed

	* correct.c, proto.h, term.c: Rename move() to imove() and erase()
	  to ierase() to avoid name clashes with the standard (ncurses)
	  versions.

2002-05-19 18:04  ed

	* Makefile, languages/altamer/Makefile,
	  languages/american/Makefile, languages/british/Makefile: Changed
	  all 'ln' commands in language makefiles to use the macro SYMLINK,
	  which can be passed down from the top-level makefile.  (Part of
	  SuSE's 3.2.06-languages.patch, generalized.)

2002-05-19 17:49  ed

	* parse.y: is_64_flag(): new function, to incorporate SuSE's patch
	  to allow more flag characters than just A-Za-z.

2002-05-19 17:32  ed

	* local.h.samp: Added comment mentioning MANFFEXT.

2002-05-19 17:19  ed

	* Makefile, test_deformatters, languages/test_dict/Makefile,
	  languages/test_dict/gen_wordlist,
	  languages/test_dict/test_dict.aff,
	  test_data/test_deformatters.html.expected,
	  test_data/test_deformatters.in,
	  test_data/test_deformatters.nroff.expected,
	  test_data/test_deformatters.ordinary.expected,
	  test_data/test_deformatters.tex.expected: Added a test case for
	  ispell's various deformatters.  This involved creating a new
	  dictionary, test_dict, which contains only a few words.  For
	  whatever reason this new test_dict is not included in
	  'all-languages', it has its own Makefile targets which are used
	  only by 'test' and 'clean'.

	  With that dictionary available, test_deformatters runs ispell
	  four times on the sample input test_data/test_deformatters.in,
	  with four different deformatters (TeX, nroff, HTML and
	  'ordinary'), and compares the results to expected values.

	  Also changed 'make test' to clean up after itself, since the
	  ispell binary left over from testing installation is no use.

2002-05-19 16:59  ed

	* ispell.1X, ispell.c: Added -o flag to force ordinary text (no
	  nroff deformatting etc.).  This is necessary to override
	  DEFTEXFLAG.

2002-05-19 15:53  ed

	* test_inst: Whitespace fixes.

2001-11-08 13:25  ed

	* Makefile: The  'clean' and 'test' targets are passed down to the
	  languages/ directory.

2001-11-07 18:59  ed

	* languages/: Makefile, fix8bit.c, test_fix8bit,
	  test_data/ff.0.7.ref, test_data/ff.0.8.ref, test_data/ff.0.in,
	  test_data/hex.0.7.ref, test_data/hex.0.8.ref, test_data/hex.0.in,
	  test_data/hex.1.7.ref, test_data/hex.1.8.ref, test_data/hex.1.in,
	  test_data/hex.10.7.ref, test_data/hex.10.8.ref,
	  test_data/hex.10.in, test_data/hex.11.7.ref,
	  test_data/hex.11.8.ref, test_data/hex.11.in,
	  test_data/hex.12.7.ref, test_data/hex.12.8.ref,
	  test_data/hex.12.in, test_data/hex.13.7.ref,
	  test_data/hex.13.8.ref, test_data/hex.13.in,
	  test_data/hex.14.7.ref, test_data/hex.14.8.ref,
	  test_data/hex.14.in, test_data/hex.15.7.ref,
	  test_data/hex.15.8.ref, test_data/hex.15.in,
	  test_data/hex.16.7.ref, test_data/hex.16.8.ref,
	  test_data/hex.16.in, test_data/hex.2.7.ref,
	  test_data/hex.2.8.ref, test_data/hex.2.in, test_data/hex.3.7.ref,
	  test_data/hex.3.8.ref, test_data/hex.3.in, test_data/hex.4.7.ref,
	  test_data/hex.4.8.ref, test_data/hex.4.in, test_data/hex.5.7.ref,
	  test_data/hex.5.8.ref, test_data/hex.5.in, test_data/hex.6.7.ref,
	  test_data/hex.6.8.ref, test_data/hex.6.in, test_data/hex.7.7.ref,
	  test_data/hex.7.8.ref, test_data/hex.7.in, test_data/hex.8.7.ref,
	  test_data/hex.8.8.ref, test_data/hex.8.in, test_data/hex.9.7.ref,
	  test_data/hex.9.8.ref, test_data/hex.9.in,
	  test_data/hex_truncate.0.7.ref, test_data/hex_truncate.0.8.ref,
	  test_data/hex_truncate.0.in, test_data/hex_truncate.1.7.ref,
	  test_data/hex_truncate.1.8.ref, test_data/hex_truncate.1.in,
	  test_data/hex_truncate.2.7.ref, test_data/hex_truncate.2.8.ref,
	  test_data/hex_truncate.2.in, test_data/hex_truncate.3.7.ref,
	  test_data/hex_truncate.3.8.ref, test_data/hex_truncate.3.in,
	  test_data/hex_truncate.4.7.ref, test_data/hex_truncate.4.8.ref,
	  test_data/hex_truncate.4.in, test_data/octal_truncate.0.7.ref,
	  test_data/octal_truncate.0.8.ref, test_data/octal_truncate.0.in,
	  test_data/octal_truncate.1.7.ref,
	  test_data/octal_truncate.1.8.ref, test_data/octal_truncate.1.in,
	  test_data/octal_truncate.2.7.ref,
	  test_data/octal_truncate.2.8.ref, test_data/octal_truncate.2.in,
	  test_data/octal_truncate.3.7.ref,
	  test_data/octal_truncate.3.8.ref, test_data/octal_truncate.3.in,
	  test_data/octal_truncate.4.7.ref,
	  test_data/octal_truncate.4.8.ref, test_data/octal_truncate.4.in,
	  test_data/rand.7.ref, test_data/rand.8.ref,
	  test_data/rand.h.7.ref, test_data/rand.h.8.ref,
	  test_data/rand.h.in, test_data/rand.hu.7.ref,
	  test_data/rand.hu.8.ref, test_data/rand.hu.in,
	  test_data/rand.hux.7.ref, test_data/rand.hux.8.ref,
	  test_data/rand.hux.in, test_data/rand.hx.7.ref,
	  test_data/rand.hx.8.ref, test_data/rand.hx.in, test_data/rand.in,
	  test_data/rand.u.7.ref, test_data/rand.u.8.ref,
	  test_data/rand.u.in, test_data/rand.ux.7.ref,
	  test_data/rand.ux.8.ref, test_data/rand.ux.in,
	  test_data/rand.x.7.ref, test_data/rand.x.8.ref,
	  test_data/rand.x.in, test_data/rand_gen: Rewrite of fix8bit.c
	  prompted by SuSE's ispell-3.2.06-languages.patch.  I wanted to
	  make sure the patch wouldn't break anything, so I wrote a test
	  suite, but doing that I found lots of other things that were
	  wrong, so I started trying to fix those...

	  Makefile: fixed dependencies for fix8bit, added 'test' target.
	  The test suite checks fix8bit's pushback routines, runs
	  test_fix8bit (see below) and checks a couple of additional
	  properties: fix8bit -8 | fix8bit -7 == cat; fix8bit -7 | fix8bit
	  -7 == fix8bit -7.

	  fix8bit.c: rewrote to8bit() to better handle cases when the
	  backslashed sequence turns out to be illegal.  The initial
	  backslash is printed and the remaining characters are pushed back
	  to be read again.  This means that for example \\x41 will print
	  as \A, in the same way that !\x41 produces !A.  It also handles
	  escape sequences cut off by EOF properly (again they are printed
	  unchanged).  This uses a mini pushback library which has a test
	  suite if you give main() the argument --test-pushback.  Also
	  fixed the original problem with hex sequences being miscomputed,
	  which SuSE wrote the patch for.  Added a warning if the input
	  already contains 8-bit chars (that would stop -8 | -7 being
	  identity).

	  test_fix8bit: new file.  This is a Perl script to run fix8bit -7
	  and fix8bit -8 on every input file in test_data/ and check the
	  results against the expected results also in that directory.

	  test_data/: new directory.  Contains test cases, some written by
	  hand and some randomly generated by rand_gen.  rand_gen tries to
	  make 'well behaved' input that doesn't muck up fix8bit -8 |
	  fix8bit -7 or other commands - but there are three flags you can
	  use to tell it not to.  The random test cases have not been
	  checked by hand, so they should be used in addition to
	  human-written ones.

2001-10-30 14:41  ed

	* config.X: Incorporating more changes from SuSE's
	  ispell-3.2.06-config.patch.

	  Raised the arbitrary limits INPUTWORDLEN, MAXAFFIXLEN and
	  SORTPERSONAL, but avoiding the silly powers of two which SuSE
	  chose.

	  The temporary directory for sorting has changed from /usr/tmp/
	  (not FHS approved) to /tmp/; we could use /var/tmp/ but there is
	  no particular need for temporary files to be preserved from one
	  run of sort to the next.  The FHS doesn't give any suggestion
	  that /var/tmp/ can be expected to have more space free than plain
	  /tmp/.

	  Spelling fixes - sheesh, if the same ironic problems applied to
	  other software then gcc wouldn't compile and gzip's tarball would
	  be corrupt :-).

2001-10-30 14:11  ed

	* config.X: Switched the default TERMLIB to ncurses from termcap.
	  Termcap is more or less obsolete in favour of ncurses.

2001-10-16 17:55  ed

	* Makefile, README, config.X, ispell.1X, test_inst,
	  languages/altamer/Makefile, languages/american/Makefile,
	  languages/british/Makefile, languages/english/Makefile: Merged
	  some more of SuSE's ispell-3.2.06-config.patch: the wordlist is
	  in /usr/share/dict/ rather than /usr/dict/.  I changed this in
	  all documentation, but the manual page ought to use the WORDS
	  configuration variable rather than anything hardcoded.  Extended
	  test_inst to check filenames mentioned in manual pages.

2001-10-16 16:19  ed

	* Makefile: Merged some of SuSE's ispell-3.2.06-config.patch:
	  local.h depends on local.h.samp.

2001-10-16 16:14  ed

	* Makefile, Makekit, README, config.X, ispell.1X, ispell.4,
	  ispell.5X, ispell.c, local.h.samp, test_inst,
	  deformatters/Makefile, languages/altamer/Makefile,
	  languages/american/Makefile, languages/british/Makefile,
	  languages/dansk/Makefile, languages/deutsch/Makefile,
	  languages/english/Makefile, languages/english/english.4l,
	  languages/english/english.5X, languages/francais/Makefile,
	  languages/nederlands/Makefile, languages/norsk/Makefile,
	  languages/svenska/Makefile: Merged some more patches from SuSE's
	  ispell-3.2.06 package: parts of ispell-3.2.06-languages.patch.
	  This is to make the manual pages have the correct section (file
	  formats go in section 5 not 4) and have correct cross-references
	  (no xref to tib (1) or spell (1), which are not installed on
	  SuSE).  I parameterized these changes as things in config.X.

	  Makefile: Fixed bug creating directories by using mkdir -p
	  instead of plain mkdir.  But -p isn't supported on all Unixes, so
	  this is parameterized as the MKDIR variable.	Perhaps this should
	  be a variable in config.X rather than the Makefile itself - but I
	  don't know how to do that.  Added ispell.5 as a target, it is
	  generated from ispell.5X by applying substitutions.  This
	  replaces the old ispell.4 which was installed as-is.

	  Two new substitution variables TIB_XREF and SPELL_XREF are used
	  in manual pages for making cross-references to tib and spell.
	  They should be defined as a fragment of troff code (ie, text) to
	  be included in manual pages.	For example, TIB_XREF might be
	  defined as 'tib (1)'.

	  The section for file format manual pages is controlled by
	  MANFFSECT, previously it was hardcoded as 4.	MANFFDIR and
	  MANFFEXT replace MAN4DIR and MAN4EXT respectively.

	  The target install-deformatters no longer calls a recursive
	  Makefile inside the deformatters/ directory.	Instead the
	  commands to build these two programs are included in the
	  top-level Makefile.  Ditto clean-deformatters.

	  To test the changes to installation, there is a new program
	  test_inst which sets some variables in local.h, runs make
	  install, and checks the installed files.  It concentrates on
	  making sure the manual pages are okay, reading them and checking
	  cross-references.  test_inst is now the last target of 'make
	  test'.

	  Dropped support for manual section 1L (local); now the executable
	  commands go in plain old section 1.  It could be added back in
	  using a similar mechanism to the section 4 / 5 choice if it is
	  still needed.

	  The Makefiles for the individual languages have also been changed
	  to use MKDIR, and bugfixed to work when LIBDIR is a relative
	  path.  Previously, Makefiles in subdirectories would do 'cd
	  $$LIBDIR', but that will fail if LIBDIR is something like
	  '../install'.  They now go through some contortions to keep
	  working whether LIBDIR is relative or absolute.  There is a lot
	  of duplicated code across the language directories, it would be
	  better factored out.	The english Makefile now builds the
	  english.5 manual page by applying substitutions from english.5X,
	  and its variants (altamer, american and british) now call
	  english/Makefile to make this page and for 'make clean'.

2001-10-08 13:09  ed

	* Makefile, ispell.1X, languages/dansk/Makefile,
	  languages/deutsch/Makefile, languages/english/msgs.h,
	  languages/espanol/Makefile, languages/francais/Makefile,
	  languages/nederlands/Makefile, languages/norsk/Makefile,
	  languages/portugues/Makefile, languages/svenska/Makefile: Purge
	  all traces of 'sq' and 'unsq'.  (Apart from the log entries :-))

2001-10-08 13:00  ed

	* Makefile: Parameterize ln(1) to $(SYMLINK).

2001-10-05 15:37  ed

	* ispell.1X: Documentation fix from SuSE
	  (ispell-3.2.06-config.patch).

2001-10-05 15:28  ed

	* Makefile: Make sure that realclean removes all files generated:
	  remove config.sh and test output.

2001-10-05 15:24  ed

	* exp_table.c, exp_table.h, expansion_diff, test_data/e1.expected,
	  test_data/e2.expected, test_data/e3.expected,
	  test_data/e4.expected, test_data/e5.expected, test_data/input,
	  test_data/rand_gen: Adding new files for 3.2.06.epa1.

2001-10-05 15:22  ed

	* Magiclines, Makefile, Makekit, Makepatch, README, buildhash.c,
	  config.X, correct.c, defmt.c, dump.c, fields.3, fields.c,
	  fields.h, findaffix.X, good.c, hash.c, icombine.c, ijoin.c,
	  ispell.1X, ispell.4, ispell.c, ispell.h, iwhich, local.h.samp,
	  lookup.c, makedent.c, makedict.sh, munchlist.X, parse.y, proto.h,
	  subset.X, term.c, tgood.c, tree.c, tryaffix.X, version.h,
	  xgets.c, zapdups.X, deformatters/Makefile,
	  deformatters/deformat-c.c, deformatters/deformat-sh.c,
	  languages/Makefile, languages/fix8bit.c,
	  languages/altamer/Makefile, languages/american/Makefile,
	  languages/british/Makefile, languages/dansk/Makefile,
	  languages/deutsch/Makefile, languages/english/Makefile,
	  languages/english/english.4l, languages/english/english.aff,
	  languages/english/msgs.h, languages/espanol/Makefile,
	  languages/francais/Makefile, languages/nederlands/Makefile,
	  languages/norsk/Makefile, languages/portugues/Makefile,
	  languages/svenska/Makefile: Imported 3.2.06.epa1 release.  This
	  was previously developed using sporadic RCS for certain files,
	  but I'm not really bothered about rolling back beyond this
	  release.

2001-10-05 15:13  ed

	* Contributors, Magiclines, Makefile, Makekit, Makepatch, README,
	  WISHES, buildhash.c, config.X, correct.c, defmt.c, dump.c,
	  fields.3, fields.c, fields.h, findaffix.X, good.c, hash.c,
	  icombine.c, ijoin.c, ispell.1X, ispell.4, ispell.c, ispell.h,
	  iwhich, local.h.samp, lookup.c, makedent.c, makedict.sh,
	  munchlist.X, parse.y, proto.h, subset.X, term.c, tgood.c, tree.c,
	  tryaffix.X, version.h, xgets.c, zapdups.X, addons/xspell.shar,
	  addons/nextispell/Makefile, addons/nextispell/README,
	  addons/nextispell/configure,
	  addons/nextispell/configure.h.template,
	  addons/nextispell/configureTeX, addons/nextispell/nextispell.m,
	  addons/nextispell/services.template, deformatters/Makefile,
	  deformatters/README, deformatters/deformat-c.c,
	  deformatters/deformat-sh.c, languages/Makefile, languages/Where,
	  languages/fix8bit.c, languages/altamer/Makefile,
	  languages/american/Makefile, languages/british/Makefile,
	  languages/dansk/Makefile, languages/deutsch/Makefile,
	  languages/english/Makefile, languages/english/altamer.0,
	  languages/english/altamer.1, languages/english/altamer.2,
	  languages/english/american.0, languages/english/american.1,
	  languages/english/american.2, languages/english/british.0,
	  languages/english/british.1, languages/english/british.2,
	  languages/english/english.0, languages/english/english.1,
	  languages/english/english.2, languages/english/english.3,
	  languages/english/english.4l, languages/english/english.aff,
	  languages/english/msgs.h, languages/espanol/Makefile,
	  languages/francais/Makefile, languages/nederlands/Makefile,
	  languages/norsk/Makefile, languages/portugues/Makefile,
	  languages/svenska/Makefile: Importing official 3.2.06 release
	  into CVS.

2001-10-05 15:13  ed

	* Contributors, Magiclines, Makefile, Makekit, Makepatch, README,
	  WISHES, buildhash.c, config.X, correct.c, defmt.c, dump.c,
	  fields.3, fields.c, fields.h, findaffix.X, good.c, hash.c,
	  icombine.c, ijoin.c, ispell.1X, ispell.4, ispell.c, ispell.h,
	  iwhich, local.h.samp, lookup.c, makedent.c, makedict.sh,
	  munchlist.X, parse.y, proto.h, subset.X, term.c, tgood.c, tree.c,
	  tryaffix.X, version.h, xgets.c, zapdups.X, addons/xspell.shar,
	  addons/nextispell/Makefile, addons/nextispell/README,
	  addons/nextispell/configure,
	  addons/nextispell/configure.h.template,
	  addons/nextispell/configureTeX, addons/nextispell/nextispell.m,
	  addons/nextispell/services.template, deformatters/Makefile,
	  deformatters/README, deformatters/deformat-c.c,
	  deformatters/deformat-sh.c, languages/Makefile, languages/Where,
	  languages/fix8bit.c, languages/altamer/Makefile,
	  languages/american/Makefile, languages/british/Makefile,
	  languages/dansk/Makefile, languages/deutsch/Makefile,
	  languages/english/Makefile, languages/english/altamer.0,
	  languages/english/altamer.1, languages/english/altamer.2,
	  languages/english/american.0, languages/english/american.1,
	  languages/english/american.2, languages/english/british.0,
	  languages/english/british.1, languages/english/british.2,
	  languages/english/english.0, languages/english/english.1,
	  languages/english/english.2, languages/english/english.3,
	  languages/english/english.4l, languages/english/english.aff,
	  languages/english/msgs.h, languages/espanol/Makefile,
	  languages/francais/Makefile, languages/nederlands/Makefile,
	  languages/norsk/Makefile, languages/portugues/Makefile,
	  languages/svenska/Makefile: Initial revision

