* Re: bug#13026: Assume POSIX 1003.1-1988 or later for signal.h.
[not found] ` <50C2A8B9.3030205@cs.ucla.edu>
@ 2012-12-08 10:13 ` Eli Zaretskii
2012-12-08 17:34 ` Paul Eggert
0 siblings, 1 reply; 3+ messages in thread
From: Eli Zaretskii @ 2012-12-08 10:13 UTC (permalink / raw)
To: Paul Eggert; +Cc: emacs-devel
> Date: Fri, 07 Dec 2012 18:40:57 -0800
> From: Paul Eggert <eggert@cs.ucla.edu>
>
> As an aside, it'd be nicer if lib/Makefile were computed more
> automatically from lib/gnulib.mk on Microsoft platforms, so
> that this sort of thing wouldn't require manual editing.
That requires Windows to use the Posix configure script. Doing that
is a large job that is already under way, albeit slowly. It is a
large job because:
. neither configure.ac nor the various Makefile.in files, in
particular src/Makefile.in, currently support the MS-Windows
build, which needs some special options be used and some special
commands to be run, as seen in the makefile.w32-in files;
. the nt/ subdirectory and the parts of the build there are not
covered at all by configure.ac and Makefile.in files;
. quite a few of the tests done by configure need to be overridden,
because the necessary features are implemented only in Emacs's own
source, not in system headers and libraries; and
. configure.ac is set up to produce executables statically linked
against available optional libraries, whereas the Windows build
loads those libraries dynamically at run time (this is a minor
annoyance easily fixed in configure.ac, but there are a lot of
such annoyances, and more are added as we go).
Volunteers to help in this job are welcome.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: bug#13026: Assume POSIX 1003.1-1988 or later for signal.h.
2012-12-08 10:13 ` bug#13026: Assume POSIX 1003.1-1988 or later for signal.h Eli Zaretskii
@ 2012-12-08 17:34 ` Paul Eggert
2012-12-08 18:45 ` Gnulib automation on Windows (was: bug#13026: Assume POSIX 1003.1-1988 or later for signal.h.) Eli Zaretskii
0 siblings, 1 reply; 3+ messages in thread
From: Paul Eggert @ 2012-12-08 17:34 UTC (permalink / raw)
To: Eli Zaretskii; +Cc: emacs-devel
On 12/08/2012 02:13 AM, Eli Zaretskii wrote:
>> Date: Fri, 07 Dec 2012 18:40:57 -0800
>> > From: Paul Eggert <eggert@cs.ucla.edu>
>> >
>> > As an aside, it'd be nicer if lib/Makefile were computed more
>> > automatically from lib/gnulib.mk on Microsoft platforms, so
>> > that this sort of thing wouldn't require manual editing.
> That requires Windows to use the Posix configure script.
I was thinking of something more modest, namely to change
nt/configure.bat to copy the needed rules from
lib/gnulib.mk to lib/Makefile. This step wouldn't
all need to be done in configure.bat; it could be
precomputed on a POSIXish platform, if the Windows tools
aren't up to doing what 'sed' can do. Ideally lib/makefile.w32-in
could just say "include gnulib.mk" but it might need
to include a modified version of gnulib.mk instead.
But if the plan is to port 'configure' to Windows perhaps
this idea would be a distraction.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Gnulib automation on Windows (was: bug#13026: Assume POSIX 1003.1-1988 or later for signal.h.)
2012-12-08 17:34 ` Paul Eggert
@ 2012-12-08 18:45 ` Eli Zaretskii
0 siblings, 0 replies; 3+ messages in thread
From: Eli Zaretskii @ 2012-12-08 18:45 UTC (permalink / raw)
To: Paul Eggert; +Cc: emacs-devel
> Date: Sat, 08 Dec 2012 09:34:05 -0800
> From: Paul Eggert <eggert@cs.ucla.edu>
> CC: emacs-devel@gnu.org
>
> On 12/08/2012 02:13 AM, Eli Zaretskii wrote:
> >> Date: Fri, 07 Dec 2012 18:40:57 -0800
> >> > From: Paul Eggert <eggert@cs.ucla.edu>
> >> >
> >> > As an aside, it'd be nicer if lib/Makefile were computed more
> >> > automatically from lib/gnulib.mk on Microsoft platforms, so
> >> > that this sort of thing wouldn't require manual editing.
>
> > That requires Windows to use the Posix configure script.
>
> I was thinking of something more modest, namely to change
> nt/configure.bat to copy the needed rules from
> lib/gnulib.mk to lib/Makefile.
I don't see how this could work: gnulib.mk uses Unixy shell features,
so running "make" will need such a shell.
> This step wouldn't all need to be done in configure.bat; it could be
> precomputed on a POSIXish platform, if the Windows tools aren't up
> to doing what 'sed' can do. Ideally lib/makefile.w32-in could just
> say "include gnulib.mk" but it might need to include a modified
> version of gnulib.mk instead.
Sed is not the main problem here. Shell commands are. I don't know
how to automatically convert Bourne shell commands to commands cmd.exe
can grok. It might be possible using something like msdos/sed*.inp,
but those Sed scripts would then need the kind of maintenance effort
we are trying to avoid here.
> But if the plan is to port 'configure' to Windows perhaps
> this idea would be a distraction.
That's what I think should be done, if we want the maintenance burden
for the Windows configury to become significantly lower.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2012-12-08 18:45 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <50B7062F.1010901@cs.ucla.edu>
[not found] ` <83sj7suxdn.fsf@gnu.org>
[not found] ` <50B7D5BC.5060606@cs.ucla.edu>
[not found] ` <83ehjbv5cc.fsf@gnu.org>
[not found] ` <50BC0104.3070303@cs.ucla.edu>
[not found] ` <83k3sxr5j6.fsf@gnu.org>
[not found] ` <50C2A8B9.3030205@cs.ucla.edu>
2012-12-08 10:13 ` bug#13026: Assume POSIX 1003.1-1988 or later for signal.h Eli Zaretskii
2012-12-08 17:34 ` Paul Eggert
2012-12-08 18:45 ` Gnulib automation on Windows (was: bug#13026: Assume POSIX 1003.1-1988 or later for signal.h.) Eli Zaretskii
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/emacs.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).