RFC: parrot_config over pkg-config for dynext build configuration
plobsing at gmail.com
Wed Nov 10 16:08:12 UTC 2010
On Wed, Nov 10, 2010 at 10:12 AM, Gerd Pokorra <gp at zimt.uni-siegen.de> wrote:
> I think there is no need to make mention of "pkg-config" in the
> documentation. I believe it is better to have an example of
> "parrot_config" there. But I think not to note "pkg-config" in the PPD
> is enough. Some linux distributions like to have the file "parrot.pc".
> -- Gerd
Here's why I want to eliminate parrot.pc:
1) parrot.pc is not useful to anyone at the current time (TT #1842)
2) fixing it to be useful would violate the letter and the objectives
of PDD30 (TT #735)
3) it is a duplicate configuration mechanism. it's existence muddies
the waters when somebody wants to know how to build against parrot.
everyone should always be using parrot_config for anything parrot
If some linux distro wants to have a parrot.pc, they are free to pull
the values out of parrot_config and generate it themselves. It is
likely to be disused, fall out of date, and generally be wrong; just
like the current situation.
> Am Mittwoch, den 10.11.2010, 05:26 +0100 schrieb Peter Lobsinger:
>> PDD30 and pkg-config don't work well together leading to the problems
>> described in TT #1842. In my opinion, the purpose of pkg-config
>> doesn't really agree with the goal of multiple installed versions.
>> Further, it is not likely to be installed on many systems we are
>> parrot_config should contain any parrot configuration information,
>> including whatever is contained in parrot.pc. It is also already the
>> recommended way for languages to build themselves for a particular
>> parrot (pdd30). Rakudo already uses parrot_config for this purpose.
>> Therefore I wish to deprecate parrot.pc and rewrite the appropriate
>> sections of docs/embed.pod to use parrot_config in stead.
More information about the parrot-dev