[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: No python in koji



On Fri, Aug 17, 2007 at 02:47:12PM -0500, Jason L Tibbitts III wrote:
> >>>>> "AT" == Axel Thimm <Axel Thimm ATrpms net> writes:
> 
> AT> No, I guess it won't, it only errored out on the Requires. But I
> AT> still feel this is very thin ice we're walking on.
> 
> The method is required for everything else that wants to encode some
> information into the spec which is gathered by running something which
> is BuildRequire:d.  All Ruby and PHP packages, for example.
> 
> AT> for another we preclude many use cases of %() in specfiles,
> AT> e.g. anything that will break rpm's specfile parses if %() returns
> AT> %{nil}.
> 
> You define your macros in such a way that they are provided
> meaningless but syntactically correct values in the case that the
> necessary executables aren't there.

So all the say PHP stuff like

Requires:     php(zend-abi) = %{php_zend_api}
Requires:     php(api) = %{php_core_api}
Requires:     php-api = %{php_apiver}

have such failsafe definitions in the rhs macros? Obviously they do.

I understand the chicken-and-egg situation with %() in BRs, so BRs
should be kind of macro-less, but the rest should be able to use
anything the BRs provide w/o having to go through loops.

I think the workaround of requiring such careful design of macros is
more work than fixing koji.</says_the_man_wo_providing_the_patch>

Anyway my packages are in the build queue, if they get out
successfully (which they will, I hope) I'm a happy man, what else does
a man want of his life? ;)
-- 
Axel.Thimm at ATrpms.net

Attachment: pgpDnITRz49gJ.pgp
Description: PGP signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]