Add FreeBSD* to configure.in for --enable-shared

[jwpye]

BPO	1044395
Nosy	@loewis, @hyeshik
Files	addfreebsd: configure.in patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/hyeshik'
closed_at = <Date 2004-10-26.09:55:26.000>
created_at = <Date 2004-10-11.07:40:40.000>
labels = ['build']
title = 'Add FreeBSD* to configure.in for --enable-shared'
updated_at = <Date 2004-10-26.09:55:26.000>
user = 'https://bugs.python.org/jwpye'

bugs.python.org fields:

activity = <Date 2004-10-26.09:55:26.000>
actor = 'hyeshik.chang'
assignee = 'hyeshik.chang'
closed = True
closed_date = None
closer = None
components = ['Build']
creation = <Date 2004-10-11.07:40:40.000>
creator = 'jwpye'
dependencies = []
files = ['6296']
hgrepos = []
issue_num = 1044395
keywords = ['patch']
message_count = 13.0
messages = ['47054', '47055', '47056', '47057', '47058', '47059', '47060', '47061', '47062', '47063', '47064', '47065', '47066']
nosy_count = 4.0
nosy_names = ['loewis', 'hyeshik.chang', 'jiwon', 'jwpye']
pr_nums = []
priority = 'high'
resolution = 'fixed'
stage = None
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue1044395'
versions = ['Python 2.4']

[jwpye]

Add FreeBSD to the same class as Linux, GNU, and NetBSD
so that the shared library will be built and installed
on FreeBSD systems when --enabled-shared is specified.

I didn't regenerate the configure script as I wasn't
sure what version of autoconf would be desired. I'll
leave that up to the committer, if accepted.

[loewis]

Logged In: YES
user_id=21627

Looks good to me.

[jiwon]

Logged In: YES
user_id=595483

Adding a patch that fixes configure (and configure.in). Used
autoconf2.59 as is specified in previous configure script.

Patch is at
http://seojiwon.dnip.net:8000/~jiwon/configure.patch

[hyeshik]

Logged In: YES
user_id=55188

FreeBSD port (their packaging system) already has
a scheme for shared library.

http://www.freebsd.org/cgi/cvsweb.cgi/ports/lang/python/files/patch-configure?rev=1.10&content-type=text/x-cvsweb-markup

It would be good if Python's setting become simliar
to FreeBSD port's.

[jiwon]

Logged In: YES
user_id=595483

http://seojiwon.dnip.net:8000/~jiwon/configure.patch2

Here's a patch conforming to FreeBSD port scheme. (I think)
However, for "ac_cv_pthread_system_supported" problem, I
think it's better patched in port system, isn't it?

[jwpye]

Logged In: YES
user_id=1044177

This might be a question for the port maintainer, but what's
the point of cutting the .0 off the SOVERSION? Having it
there hasn't caused any trouble for my CVS build(yet?)..

[hyeshik]

Logged In: YES
user_id=55188

For jiwon's patch:

You don't need to post a patch to configure, only
configure.in is enough. For line 1347,1354, I think we
don't need to keep using not $CC but 'cc' as linker on
OpenBSD. AFAIK, OpenBSD uses gcc as their compiler set and
there're no alternatives available. So, it'll not need to
have a special case for OpenBSD.

And, ac_cv_pthread_system_supported stuff is to fix Python's
pthread scope behavior that is not appropriate on FreeBSD.
(In FreeBSD, system scope threads are very expensive than
process scope threads and the former has lots of limitations
on resource usages.) I think it can go into Python src as a
ifdef'ed block, I will post a patch in a new bug item soon.

To jwpye:

FreeBSD's runtime linker system can't handle so versions
with dot(s). You can link againt shared objects with dotted
so versions if you explicitly specified rpath on linking.
But -lpython2.3.1.0 may not be found by ldconfig(1) nor ld
-shared.
Hence, it tends to make KDE and/or GNOME applications using
python difficult to compile.

[jiwon]

Logged In: YES
user_id=595483

Okay.
Here's the patch reflection your suggestion.

http://seojiwon.dnip.net:8000/~jiwon/configure.in

[loewis]

Logged In: YES
user_id=21627

perky, I fail to understand your comment on versioning. It
appears to suggest that the patch attached to this report
does not work. However, if it doesn't work, I'm surprised
that jwpye suggested it in the first place.

Anyway, assigning the entire issue to you for resolution.