Stupid, stupid, stupid Reply rules on this list... Forwarded message: (from me)
Boey Maun Suang wrote:
In addition, it's not much trouble at all to make a port install man pages into ${prefix}/share/man; in most cases, we either pass --mandir=${prefix}/share/man to configure or copy it ourselves. (I suspect that most of the ports that install man pages into ${prefix}/man are using configure script generated by autoconf < 2.59c, as it was only at that revision that it changed its default mandir to ${prefix}/share/man.) Consequently, I think we should stick with installing man pages into ${prefix}/share/man.
Seems very redundant to have that --mandir in each and every little Portfile. Might as well have the definition next to the default --prefix=${prefix}, so that it would automatically add --mandir=${prefix}/share/man --infodir=${prefix}/share/info ? Since that is now the mandatory location, might as well make it the default configure arguments as well...
Some ports that install into ${prefix}/man either use their own build system or plain have it hardcoded in their Makefiles. For example, MacPorts have hardcoded the location of ${prefix}/share/man in doc/Makefile, no matter what ${mandir} is ? The main reason why it didn't used to be such a big fuzz, was because of the ${prefix}/man -> share/man symlink.
--anders
On Aug 14, 2007, at 02:24, Anders F Björklund wrote:
Stupid, stupid, stupid Reply rules on this list...
The list configuration is not stupid, for the reasons outlined here: http://www.unicom.com/pw/reply-to-harmful.html
Forwarded message: (from me)
Boey Maun Suang wrote:
In addition, it's not much trouble at all to make a port install man pages into ${prefix}/share/man; in most cases, we either pass --mandir=${prefix}/share/man to configure or copy it ourselves. (I suspect that most of the ports that install man pages into $ {prefix}/man are using configure script generated by autoconf < 2.59c, as it was only at that revision that it changed its default mandir to ${prefix}/share/man.) Consequently, I think we should stick with installing man pages into ${prefix}/share/man.
Seems very redundant to have that --mandir in each and every little Portfile. Might as well have the definition next to the default --prefix=${prefix}, so that it would automatically add -- mandir=${prefix}/share/man --infodir=${prefix}/share/info ? Since that is now the mandatory location, might as well make it the default configure arguments as well...
Adding that to the configure.pre_args would probably break all the ports whose configure scripts don't recognize the --mandir argument, wouldn't it? I hate MacPorts base changes that break whole swathes of ports. I feel that people should feel free to make such changes to base, so long as they also modify all affected ports so they do not break as a result. If that is too cumbersome, then IMHO the change to base should not be made. For example, making mtree violations fatal errors broke lots of ports. That should not have been done.
Some ports that install into ${prefix}/man either use their own build system or plain have it hardcoded in their Makefiles. For example, MacPorts have hardcoded the location of ${prefix}/share/ man in doc/Makefile, no matter what ${mandir} is ? The main reason why it didn't used to be such a big fuzz, was because of the $ {prefix}/man -> share/man symlink.
And why is it now a big deal to track these mtree violations? I still have said symlink.
Ryan Schmidt wrote:
Stupid, stupid, stupid Reply rules on this list...
The list configuration is not stupid, for the reasons outlined here:
As long as you can live with a few responses not making it to the list, due to me hitting the wrong key/button then everything is fine with me. :-)
ding that to the configure.pre_args would probably break all the ports whose configure scripts don't recognize the --mandir argument, wouldn't it? I hate MacPorts base changes that break whole swathes of ports. I feel that people should feel free to make such changes to base, so long as they also modify all affected ports so they do not break as a result. If that is too cumbersome, then IMHO the change to base should not be made. For example, making mtree violations fatal errors broke lots of ports. That should not have been done.
Most likely it would break ports not using regular autoconf configure, yes. It mostly works for the %configure macro in RPM, but certainly not everywhere.
Some ports that install into ${prefix}/man either use their own build system or plain have it hardcoded in their Makefiles. For example, MacPorts have hardcoded the location of ${prefix}/share/man in doc/Makefile, no matter what ${mandir} is ? The main reason why it didn't used to be such a big fuzz, was because of the ${prefix}/man -> share/man symlink.
And why is it now a big deal to track these mtree violations? I still have said symlink.
I don't know either, earlier changes were delayed due to DarwinPorts->MacPorts, and now they are being delayed due to man->share/man. Both petty, in my book :-P --anders
On Aug 14, 2007, at 02:48, Anders F Björklund wrote:
Ryan Schmidt wrote:
Some ports that install into ${prefix}/man either use their own build system or plain have it hardcoded in their Makefiles. For example, MacPorts have hardcoded the location of ${prefix}/share/ man in doc/Makefile, no matter what ${mandir} is ? The main reason why it didn't used to be such a big fuzz, was because of the ${prefix}/man -> share/man symlink.
And why is it now a big deal to track these mtree violations? I still have said symlink.
I don't know either, earlier changes were delayed due to DarwinPorts->MacPorts, and now they are being delayed due to man->share/man. Both petty, in my book :-P
What is being delayed?
Ryan Schmidt wrote:
I don't know either, earlier changes were delayed due to DarwinPorts->MacPorts, and now they are being delayed due to man->share/man. Both petty, in my book :-P
What is being delayed?
New features, but maybe I'm just imagining things. I'll try to add some to trunk, once the MacPorts 1.5.11 release is out. (maybe needs a stable/dev split sometime) --anders
participants (2)
-
Anders F Björklund
-
Ryan Schmidt