rpms/poker-eval/FC-4 poker-eval.spec,1.8,1.9

Ralf Corsepius rc040203 at freenet.de
Sat Jun 10 14:26:38 UTC 2006


On Sat, 2006-06-10 at 04:49 -0700, Panu Matilainen wrote:
> On Sat, 10 Jun 2006, Christopher Stone wrote:
> 
> > On 6/10/06, Michael Schwendt <bugs.michael at gmx.net> wrote:
> >> Calm down, everyone. Ralf is right. Using %makeinstall should be
> >> discouraged actually, and, for instance, I was under the impression that
> >> we agreed on that long ago.
> >
> > No one is disagreeing with Ralf, we are only suggesting that this
> > information be added to the wiki page somewhere so new people coming
> > in can learn about these things.  I think it would be nice to have
> > some kind of documentation on the macros available to packagers
> > somewhere.  Are there any other macros that I should not be using?
> 
> How about replacing the %makeinstall macro definition with something 
> useful instead?
This doesn't work, because it would break compatibility and will have to
stay for many years, if you don't want to break compatibility.

It's just that %makeinstall isn't wrong, it's simply that most Makefile
nowadays support a superior mechanism, which renders using %makeinstall
into an anachronism/discouraged and redenders changing a functional spec
file using a functional "make DESTDIR=.. install" to using %makeinstall
a very questionable step.

>  "There's this convenient macro but you shouldn't use it" 
> seems silly to me.
Nope, it's just a case of "Think about what you are doing".

There are no 100% bullet proof rule to use %makeinstall etc. Packagers
can't avoid to check for what a package expects and should chose the way
they consider the most convenient to them. In some cases this will be
macros, in some cases it will be other steps.

Finally, in case somebody wants to write a portable spec file, ... don't
use any macros - They are all broken on some distros (comprising RH and
Fedora).

Ralf






More information about the fedora-extras-list mailing list