Michael Homer wrote: > > I'll let Hisham speak for himself, but I think the general point he's > making (and the one that's been discussed before) is that in general, > running something in between configure and make is a hack to fix > something that's been misconfigured (e.g., patching a makefile > post-generation). That sort of situation is better handled by patching > configure.ac to fix the problem at the root, which will probably keep > working between releases and be more robust. > Although I agree on this point, I also think that we should trust recipes authors a little more. In other words, I believe adding features to our recipes system, even if only usable by specific type recipes, should just be seen as a "handy" addition, which could or could not be used when necessary. It's up to the recipe writer, hence, to not misuse what we provide him with to best accomplish what he need to. For example in the sun-jdk recipe the pre_patch hook is used to run the license and unpack the package, though no patch is actually present. So I think this could be seen as a misuse of the pre_patch hook (which the docs tell should not be executed when no patches are present, but this doesn't seem to reflect reality).
Also, as you said, OOo has a very uncommon build procedure, for which I really don't see any clean way to perform it without this post_configure hook. But I'm surely not that deep into Compile internals as Hisham or any other dev is. > The other objection is that pre_configure() would only apply to > configure recipes, while all the other hooks are general, and that > type-specific hooks are a net bad. > Again, I can't honestly see why. I wouldn't want the need to be as clean and general as possible to become a too restrictive constraint. Thanks for your patience. _______________________________________________ gobolinux-devel mailing list gobolinux-devel@lists.gobolinux.org http://lists.gobolinux.org/mailman/listinfo/gobolinux-devel
