> "John" == John Poltorak <[EMAIL PROTECTED]> writes:
John> Is there any chance of including OS/2 in config.guess ?
Send patches to [EMAIL PROTECTED]
| #ifdef HAVE_SIGNAL_H
| # include
| #endif
We promote #if over #ifdef.
> "Golubev" == Golubev I N <[EMAIL PROTECTED]> writes:
Golubev> The following patch to 2.13 adds macros to perform
Golubev> recompilation avoidance. They might be used like this
Golubev> (recompilations in AC_CHECK_LIB remain)
Probably not a bad idea, but the patch is too old to be applied,
On Thu, Jan 11, 2001 at 03:27:20PM +0100, Akim Demaille wrote:
:
: | #ifdef HAVE_SIGNAL_H
: | # include
: | #endif
:
: We promote #if over #ifdef.
Why? Isn't that only the case for "boolean" defines (either 0 or 1, not
defined or not defined)? I think the Borland compiler has problems with
"
> "Lars" == Lars J Aas <[EMAIL PROTECTED]> writes:
Lars> On Thu, Jan 11, 2001 at 03:27:20PM +0100, Akim Demaille wrote:
Lars> :
Lars> : | #ifdef HAVE_SIGNAL_H : | # include : | #endif
Lars> :
Lars> : We promote #if over #ifdef.
Lars> Why?
Mostly elegance, and since the introduction of AC
On Thu, Jan 11, 2001 at 04:24:54PM +0100, Akim Demaille wrote:
: Lars> Isn't that only the case for "boolean" defines (either 0 or 1,
: Lars> not defined or not defined)? I think the Borland compiler has
: Lars> problems with "#if " if isn't defined to
: Lars> something... I'll have to check ag
Could you try this?
eval 'echo toto >/tmp/stdout 2>/tmp/stderr'
and
eval 'echo toto >/tmp/stdout 2>/tmp/stderr' >/tmp/outter
?
Thanks!
> "Lars" == Lars J Aas <[EMAIL PROTECTED]> writes:
Lars> Of course. No Autoconf/Automake-based package support building
Lars> with the Borland C++ compiler yet AFAIK.
I knew what you were to respond :) We'll see when trouble arise for real.
> "Assar" == Assar Westerlund <[EMAIL PROTECTED]> writes:
Assar> I get:
Assar> configure.in:9: error: undefined macro: AF_INET
As a matter of fact it is a warning in the sense that configure is OK.
Assar> which is apperently since A[A-Z]_ is reserved for autoconf.
Assar> But surely this sh
> "Assar" == Assar Westerlund <[EMAIL PROTECTED]> writes:
Assar> Running a configure generated from the current autoconf on
Assar> HP-UX 10.20 gives me an executable prefix '0'.
Ah!
Could you please try to run the test suite on this host? I find it
most unpleasant that this problem
> "Harlan" == Harlan Stenn <[EMAIL PROTECTED]> writes:
Harlan> Folks, It might be a local scripting problem - my current
Harlan> thought is that a strange local environmental case caused two
Harlan> builds to be directed into the same build directory.
But is the problem new?
Harlan> I'm tes
> "Assar" == Assar Westerlund <[EMAIL PROTECTED]> writes:
Assar> So it would seem to be a race-condition which would indicate
Assar> that configure is creating files in srcdir in error.
make distcheck runs the testsuite with srcdir != builddir, and srcdir
read-only. I'm checking again now,
| I found another bug in AC_FUNC_MEMCMP. It lacks `,' between 1st
| and 2nd argument to AC_TRY_RUN. On any system, configure cannot
| find a working memcmp() because the line
Aaah! How atrocious. Thanks a lot for having spotted this, it's
being applied.
Hi,
I'd like to know if autoconf has a macro which tests for some system defines
like SA_INTERRUPT or the PTHREAD_STACK_MIN
with regards,
Oliver
> "Alexandre" == Alexandre Oliva <[EMAIL PROTECTED]> writes:
Harlan> checking whether build environment is sane... yes
Harlan> /bin/sh: \272: not found
Alexandre> What if you take the quotes off "$prog" ?
Harlan> That seems to be the ticket!
Warf warf warf!
272 = 200 (8th bit) + 72 (ascii
Libtool 1.3.5 comes with it's own poor exe extension test (ltconfig).
The Autoconf macro detects the correct exe-ext, but later when libtool
is being configured, libtool doesn't inherit what configure already
has detected, except if you use a cache file. You're OK with the default
behaviour if y
If you invoke AC_LANG(C++) above AC_PROG_LIBTOOL, you will find
that $ac_ext is "c" after AC_PROG_LIBTOOL has been run, and
invoking AC_LANG(C++) again will not dispatch the necessary shell
code to fix this again since AC_LANG thinks everything is hunky
dory. You need to switch to another languag
Hello!
Please add a link to the development versions of Autoconf
(ftp://alpha.gnu.org/gnu/autoconf/) on the autoconf homepages, i.e.
http://sources.redhat.com/autoconf/ and
http://www.gnu.org/software/autoconf/
Regards,
Pavel Roskin
Akim Demaille <[EMAIL PROTECTED]> writes:
> > "Assar" == Assar Westerlund <[EMAIL PROTECTED]> writes:
> Assar> configure.in:9: error: undefined macro: AF_INET
>
> As a matter of fact it is a warning in the sense that configure is OK.
Yeah I realized the configure is ok (after a while), but s
Akim Demaille <[EMAIL PROTECTED]> writes:
> Could you please try to run the test suite on this host? I find it
> most unpleasant that this problem was not caught before. We will fix
> this problem, sure, but first, I want to make sure that the test suite
> is able to catch this. Then we will fi
$ eval 'echo toto >/tmp/stdout 2>/tmp/stderr'
$ cat /tmp/stdout
toto
$ cat /tmp/stderr
$
$ eval 'echo toto >/tmp/stdout 2>/tmp/stderr' >/tmp/outter
illegal io
$
Harlan
> Could you try this?
>
> eval 'echo toto >/tmp/stdout 2>/tmp/stderr'
>
> and
>
> eval 'echo toto >/tmp/stdout 2>/tmp/stderr
I am trying to use autoconf on Windows with the
MKS Toolkit, which consists of a Korn shell and
the usual Unix utilities.
I first had to modify config.guess and config.sub
to recognize the output of the 'uname' command of
MKS. Currently, config.guess and config.sub seem
to only recognize Cygwin
22 matches
Mail list logo
