apparently prefix code was merged without review, so i'm just now noticing these new hack patches.
+ #ifndef __APPLE__
if ((f = setmntent (mtab_file, "r")) == NULL)
turn this into a proper configure check and send it upstream. __APPLE__ checks are unacceptable.
what is this exactly ? the patch itself lacks any sort of documentation
useless patch that isnt actually applied anywhere
(In reply to comment #0)
> apparently prefix code was merged without review, so i'm just now noticing
> these new hack patches.
Interesting, I would have expected a bug for these changes. They were obviously not meant to be migrated, given alone the amount of commented out code.
> + #ifndef __APPLE__
> if ((f = setmntent (mtab_file, "r")) == NULL)
> return errno;
> + #endif
> turn this into a proper configure check and send it upstream. __APPLE__ checks
> are unacceptable.
this entire patch is no longer necessary and now dropped
> what is this exactly ? the patch itself lacks any sort of documentation
Unfortunately, the one who imported it, redid all patches, discarding the comments/explanations at the top of the patches, including upstream bug-refs.
I added the docu in the patch again now, and made it work as well. It obviously wasn't tested.
> useless patch that isnt actually applied anywhere
1. I am sorry for removing all the comments from the patches. They didn't apply straight so I redid them, but dropped the comments.
2. I didn't asked for review, as all changes a specific to prefix arches and only applied there. So I saw no change for "normal" arches, which needs to be reviewed. My fault, if I was wrong with that.
Is everything clean now, or is there still something to fix/review.
yes, you were wrong with (2). all prefix changes being merged into the tree should be opened as a bug like any other enhancement. if anything, the sheer amount of feedback/changes being made on top of your prefix commit should be enough to convince you moving forward.
the only thing left in this bug is the darwin shared lib patch. but i think Fabian has cleaned up the patch header and moved it upstream. i'm not sure if he wants anything more to be done on it in the mean time and/or here, so i'll leave bug closure to him.
I submitted the patch upstream (long ago), I can't do more, I'm affraid. The situation is similar to the FreeBSD shared libs versioning change in Gentoo. So if you (vapier) are fine with that, we can close this bug.
what i dont understand is why. the FreeBSD port is doing it only when Gentoo is the actual distro running (CHOST=...-gentoo-...) because that means we have full control over everything. but you're talking about a prefix system where Gentoo is added on and supposed to be working with the host system, so how can you change the library semantics ?
I don't change the semantics. I only change the way the libraries are symlinked to match the rest of the system. Some packages do it manually (without libtool), and have ideas that Darwin is like FreeBSD, so also should have the versioning like that. This is incorrect.
Besides that, the patch simply adds an install_name (soname in ELF terminology) that is really common practice and necessary to get correct linking by others.
I think we shouldn't be needing e2fsprogs on Darwin any more, so I'm closing this. Feel free to reopen if there's still cleanup to be done.