Greetings, As of today, the Python 2.5 port is an island. It has the correct dependencies and builds correctly, but it also has some major problems which makes it mostly useless: 1. The Python framework is not created (--enable-framework is not specified in the configure args). 2. The 'python' symlink to 'python2.5' is not created. 3. All MacPorts python libraries and applications depend on port:python24 or are part of portgroup:python24. Is there a specific reason for (1) and (2) or is this just an overlook ? More importantly, is there a way to workaround (3) and use MacPorts- managed python libraries and applications with port:python25 instead of port:python24 ? A way to ignore dependencies maybe ? (which wouldn't completely work for all ports though, the python variant of port:boost uses a hardcoded python version variable for example) -- Luc Heinrich - luc@honk-honk.com - http://www.honk-honk.com
On 7-Dec-06, at 3:51 AM, Luc Heinrich wrote:
As of today, the Python 2.5 port is an island. It has the correct dependencies and builds correctly, but it also has some major problems which makes it mostly useless:
1. The Python framework is not created (--enable-framework is not specified in the configure args). 2. The 'python' symlink to 'python2.5' is not created. 3. All MacPorts python libraries and applications depend on port:python24 or are part of portgroup:python24.
Is there a specific reason for (1) and (2) or is this just an overlook ?
More importantly, is there a way to workaround (3) and use MacPorts- managed python libraries and applications with port:python25 instead of port:python24 ? A way to ignore dependencies maybe ? (which wouldn't completely work for all ports though, the python variant of port:boost uses a hardcoded python version variable for example)
I don't know the answer to all of these issues, but I do know that Martin Costabel has been working on Python2.5 and specifically that he has built boost and visual libraries for 2.5. This work lives in the unstable tree until it has been tested enough to be moved to stable, so if you want the cutting-edge stuff, maybe checking out the work in unstable would be worthwhile. --Dethe This [the adoption of open source software] isn't just my idea, or Brazil's idea. It's the idea of our time. The complexity of our times demands it." --Gilberto Gil, Brazilian Minister of Culture
Sorry, my last message was a mistake. Martin Costabel has been working on the unstable branch of *fink*, not macports. I'm on both lists and got that mixed up. I apologize for the confusion. --Dethe Simple things should be declarative. Complex things should be procedural. --Adam Bosworth
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dethe Elza wrote:
I don't know the answer to all of these issues, but I do know that Martin Costabel has been working on Python2.5 and specifically that he has built boost and visual libraries for 2.5. This work lives in the unstable tree until it has been tested enough to be moved to stable, so if you want the cutting-edge stuff, maybe checking out the work in unstable would be worthwhile.
--Dethe
Dethe, I think you're referring to Fink here...MacPorts has no concept of "stable/unstable." - --Kevin - -- Kevin Walzer Code by Kevin http://www.codebykevin.com -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFeEN7EsLm8HXyq4sRAur0AJ4ksvtDqTuUP889qQVxcSmqpm4MewCeLc1M DYydYoxYEd9ReM9aAhTPpvo= =IqNG -----END PGP SIGNATURE-----
On Dec 7, 2006, at 03:51, Luc Heinrich wrote:
Greetings,
As of today, the Python 2.5 port is an island. It has the correct dependencies and builds correctly, but it also has some major problems which makes it mostly useless:
1. The Python framework is not created (--enable-framework is not specified in the configure args). 2. The 'python' symlink to 'python2.5' is not created. 3. All MacPorts python libraries and applications depend on port:python24 or are part of portgroup:python24.
Is there a specific reason for (1) and (2) or is this just an overlook ?
More importantly, is there a way to workaround (3) and use MacPorts- managed python libraries and applications with port:python25 instead of port:python24 ? A way to ignore dependencies maybe ? (which wouldn't completely work for all ports though, the python variant of port:boost uses a hardcoded python version variable for example)
I'd like to propose: - The 'python' port group, which requires python2.3, be renamed to 'python23'. The one port that uses it (trac) can be updated. - A new port group be created for python2.5 (named 'python') - Existing ports referencing 'python24' are updated, as they are tested, to reference the new 'python' port group. - In the future, the 'python' port group will remain pointing at the latest stable Python. I expect that most ports work fine with python 2.5 -- it may be appropriate to do a mass change instead of testing individually. Thoughts? I want to use Python 2.5! =) -landonf
On Dec 13, 2006, at 12:29 AM, Landon Fuller wrote:
I expect that most ports work fine with python 2.5 -- it may be appropriate to do a mass change instead of testing individually.
The current release of Twisted has problems with python 2.5, appearently (most of) those will be fixed for twisted 2.5. Ronald
On 13 déc. 06, at 07:13, Ronald Oussoren wrote:
The current release of Twisted has problems with python 2.5, appearently (most of) those will be fixed for twisted 2.5.
I'm trying to manually compile Boost.Python against Python 2.5 and it doesn't work either, apparently because of the changes described in PEP 353 (http://www.python.org/dev/peps/pep-0353/). Running F.Lundh 'ssizecheck' scanner on the Boost.Python sources gives a couple dozens warnings... -- Luc Heinrich - luc@honk-honk.com - http://www.honk-honk.com
On 13 déc. 06, at 00:29, Landon Fuller wrote:
Thoughts? I want to use Python 2.5! =)
Me too, but I have found other problems: *None* of the dependencies listed in the python25 Portfile are actually used. The python24 Portfile has a post-configure step which patches the Makefile to fix this. The "normal" way to correctly build the python extensions when building manually is to set some environment variables but doing the same in MacPorts does not seem to work at all, I have tried to add some configure.env and build.env but they seem to be simply ignored... :/ Moreover, the post-configure step used in python24 does not allow the bsddb extension to be correctly built, an additional patch to setup.py is apparently necessary for this (since environment variables seem to be ignored). Oh, and the --disable-tk configure option does not seem to do what it looks like it is supposed to do. -- Luc Heinrich - luc@honk-honk.com - http://www.honk-honk.com
participants (5)
-
Dethe Elza
-
Kevin Walzer
-
Landon Fuller
-
Luc Heinrich
-
Ronald Oussoren