Re: documentation error in m4 online help

[Eric Blake-1]

Hi (and pardon the nabble footer; I'm not at my usual computer),


> In terms of time investment, autoconfiscating branch-1_4 is no big deal 
> really.  I resisted the pressure until now simply because there has been
> no obvious advantage to investing that effort.  I don't mind doing it
> in order to get some concrete return.  Now that 2.0 is at least a few
> months away,  releasing 1.4.5 is certainly worthwhile in the near term.

OK, I might start looking into that.

> Before we do that, I'd like to add Make rules to automate the production
> of the web documention with gendoc.sh.

Does gnulib's Makefile.maint already do this?

> 
> Is there anything else that needs to be in the tree before making a
> 1.4.5 release?

Yes - I still haven't gotten feedback on how to fix these two coredumps:

-
http://www.nabble.com/bug+report+and+patch%3A+handling+of+chained+wrapups-t989716.html#a4331221

m4wrap(‘format(‘%s is good.’, m4wrap(‘All done!’)‘My luck’)’)

John's proposed patch solves the coredump, but keeps GNU m4 in
violation of POSIX semantics.  For this input:

m4wrap(`a')m4wrap(`b')

GNU produces ba (using LIFO semantics, like C's atexit), but
POSIX requires ab (using FIFO semantics).

-
http://www.nabble.com/dangling+pointer+bug+and+proposed+patch-t950602.html#a4330783

define(‘f’, ‘level1 $1’)f(f(f(undefine(‘f’)stuff)))

John's proposed patch solves the coredump, but I'm not sure if his
semantics are correct in every case; at any rate, his proposal is not
consistent with Solaris m4 (and in the absense of good POSIX requirements
on what to do when a currently expanding macro is then undefined, we
might as well at least be consistent with other implementations, rather
than dumping core).


As for merging stuff from branch-1_4 to HEAD, I have a list of patches
that need forward porting at the moment.  And there is still the
question of pulling in newer lib/regex.c (and other upstream files)
from gnulib is worthwhile (last time we tried it, the new version
broke on cygwin, so you reverted it just before 1.4.4; but gnulib
has had progress in the meantime).

-- 
Eric Blake

--
View this message in context: 
http://www.nabble.com/-bob%40badour.net%3A+documentation+error+in+m4+online+help--t1655609.html#a4560632
Sent from the Gnu - M4 - Bugs forum at Nabble.com.