* TODO file for GNU recode

Tell `mailto:recode-bugs@gnu.org' if you feel like volunteering for any
of these ideas, listed more or less in decreasing order of priority.
Some TODO items are implicit from received email.  See file BACKLOG.
Significant contributions require written assignments and disclaimers.

.* Before next official release
. + Memory leaks: see add_to_sequence, ALLOC, recode_{m,re}alloc}.
. + Revise everything about explodes.
. + Update NEWS for latest libiconv and latest Keld tables.
. + Remove TeX edition warnings, rephrase whatever required.
. + Complete inter-node and menu references, as necessary.
. + Resolve THANKS, which Emacs does not save correctly.
. + Check with recent versions for Autoconf, Automake, Libtool.
. + See make and configure HAVE_UNION_WAIT, see `lr /conf-wait-'.
. + Solaris cc 4.2 with option "-O" miscompiles src/libiconv.c, probably
    the function module_libiconv. The symptom is that
    "src/recode --quiet --force ascii-bs..ebcdic < COPYING > output"
    creates an output file smaller than COPYING. The only known workaround
    is to not use "-O".

.* Keld
. + recode
.  - New tables from Keld
.  - Revise diagnostics
. + Standards aspects
.  - Wait for PostScript
.   : charset in 14652
.   : transliteration in 14652
.   : API in 15435
.   : recoding/iconv APIs (->Handa)
. + Miscellaneous
.  - African tables
.  - Unicode/W3C short-comings
.   : Lithuanian

.* Particular notes

. + feuille Tamil
. + Autoconf and DOSWIN
. + New RFC 1345 tables, still some diagnostics
. + Check notes from ORA
. @ ~/libit/rmail/m4
. @ ~/libit/rmail/gettext

.@ ChangeLog

.* Charsets
. + Clean up applemac vs macintosh and ibmpc vs ibm437.
. + Implement RFC1345 more fully.
. + Generalize older charsets (latex, texte, etc.) to RFC 1345 mnemonics.
. + Support VT320 support and related character sets.  (Alain Simon)
. + Support @code{[nt]roff} diacritics.
. + Support Universite de Montreal ``accent'' convention.

.* Surfaces
. + Implement some invbits over 8, 7 or 6 bits, maybe more.
. + CR-LF, in ibmpc.c (implied by IBMPC?)
. + CR, in applemac.c (implied by Apple-Mac?)
. + Message headers [RFC 1342]
. + Mnemonic and Mnemo (maybe?) [RFC 1345]
. + Integrate -c and -g into charsets.
. + Find something for `recode -g ibmpc:ibmpc' to do what it suggests
. + Option -M (implying -i) to process MIME headers

.* Mechanics
. + Create a Python prototype of full `recode'.
. + Intercept signals to cleanup the temporary files in any event.
. + Revise --sequence=pipe thoroughly.
. + Rewrite execute_pipe_sequence for having only one child per parent.
. + Revise --force thoroughly.
. + Rewrite .l converses for lat1{asci,ltex,txte}, for --force.
. + Optimize/merge all one-to-many cases.
. + Write a better `make check', not as partial and forgiving.
. + Modify recode so I can get rid of diacrit.[ch] (elsewhere).
. + Have -l know which output charset to use (Martin Maechler, 93-12-07).
. + Take note that ->RFC1345 recodings are non reversible.

.* Notes about MSDOS port
. + I had to `touch stamp-h'.
. + @PRODUCT@ and @VERSION@ were not substituted.
. + Replace PRODUCT by PACKAGE within config_h.*.
. + One should have -DHAVE_CONFIG_H=1 (the `=1' is required).
. + charset.c:637: Maybe incorrect assignment in init_table_for_rfc1345().
. + LIBINTL.OBJ and LOCALE.OBJ were not compiled.
. + Rule to compile LIBINTL.OBJ was missing.
. + readme.tcc should become README.DOS as things go.