Bug#373643: Second IRC discussion on the state of this bug

[Loïc Minier]

05:46 < trave11er> lool: ping
05:46 < trave11er> oh noes, contentless ping
05:46  * trave11er adds some content
05:46 < trave11er> lool: it's about 373643
[...]
08:52 < lool> trave11er: yes?
08:53 -!- Lo-lan-do [~roland at AToulouse-257-1-36-105.w90-5.abo.wanadoo.fr]=
 has joined #debian-devel
08:54 -!- dgil [~dgil at 15.Red-213-98-48.staticIP.rima-tde.net] has quit [Q=
uit: That's all Folks!]
08:56 < trave11er> lool: i've seen that you were involved in 373643, can =
you explain to me what's the problem is? why can't we just hardcode DATAD=
IRNAME to 'share' in intltool's Makefile.in.in?
08:57 -!- dgil [~dgil at 15.Red-213-98-48.staticIP.rima-tde.net] has joined =
#debian-devel
08:57 < lool> trave11er: I'm commenting in the report
08:58 < lool> trave11er: certainly not no
08:58 < lool> trave11er: it should be as competent as the usptream macro
08:58 < lool> trave11er: we're talking about m4 files here which we expec=
t module maintainers to aclocal and pack into tarballs and distribute to =
other OSes
08:59 -!- POX [pox at ala.t19.ds.pwr.wroc.pl] has joined #debian-devel
09:00 < lool> trave11er: thanks for your interest on the bug though
09:00 < trave11er> lool: within intltool this variable is only mentioned =
in Makefile.in.in
09:00 < lool> trave11er: what I was planning to do but which you can do t=
o help is build a mini test suite
09:01 < trave11er> from that i conclude that it must be substituted by co=
nfigure
09:01 < lool> trave11er: I've sent a comment
09:01 < trave11er> lool: ok
09:02 < lool> trave11er: yes, the problem is that the dependency on the G=
LIB macro was not coded in the intltool macro, and I don't want to add su=
ch a dependency on another broken package (glib-gettext)
09:02 < lool> trave11er: beside, you must keep in mind that you should no=
t break configure.acs written for the old intltool.m4 (0.34)
09:03 -!- Huahua [~hua_ at 122.0.229.113] has quit [Read error: Connection r=
eset by peer]
09:04 < trave11er> lool: i can't find glib-gettext in the archive
09:04 < lool> trave11er: it's in libglib2.0-dev
09:04 < trave11er> ok
09:05 < lool> trave11er: in my comment, I pointed out one very bad design=
 of glib-gettext; it's so bad looking that no one dares fixing it upstrea=
m, and tarball maintainers are moving away of glib-gettext
09:05 < trave11er> lool: so it's basically must be set to 'share' for eve=
rything, except solaris, or something?
09:06 < lool> trave11er: something like that yes
09:08 < trave11er> lool: so ripping out bits and pieces defining and subs=
tituting DATADIRNAME from glib-gettext and pushing them into intltool.m4 =
would be an acceptable solution?
09:08 < lool> trave11er: I'm not sure
09:10 < lool> trave11er: my personal current favorite would be to 1) swit=
ch to INTLTOOL_DATADIRNAME in both Makefile.in.in and intltool.m4 2) vers=
ion the Makefile.in.in and intltool;M4 in a way that bad mixtures are avo=
ided 3) make INLTOOL_DATADIRNAME default to DATADIRNAME if set
09:10 -!- zufus [~zufus at 193.205.81.2] has joined #debian-devel
09:10 < lool> trave11er: but I wanted to write a testsuite first as I don=
't want to ship a broken macro
09:10 -!- POX [pox at ala.t19.ds.pwr.wroc.pl] has quit [Ping timeout: 480 se=
conds]
09:11 -!- heretik [~heretik at tor-irc.dnsbl.oftc.net] has quit [Quit: Clien=
t exiting]
09:11 < lool> trave11er: mind if I send the IRC log to the bug?  (or woul=
d you?)
09:11 < trave11er> go ahead

--=20
Lo=C3=AFc Minier <lool at dooz.org>