Bugzilla Bug 158730

emerging app-editors/emacs-cvs-22.0.91 fails during the install phase as
upstream have bumped the version in cvs to 22.0.92

*** Bug 158768 has been marked as a duplicate of this bug. ***

*** Bug 159014 has been marked as a duplicate of this bug. ***

Although I am on holiday...this issue needed to be fixed.  Thanks for
reporting.

Christian,

I think that what you've done is a Band-Aid rather than a root cause fix.  What
I determined is that the ebuild was building the HEAD revision from the CVS
tree as opposed to the 22.0.91 revision.  This was causing some of the files to
be named things that the ebuild didn't expect, which was causing it to die.

Looking at the CVS logs, the correct tag for the 22.0.91 version is
"EMACS_PRETEST_22_0_91".  This should be the value of the ECVS_BRANCH variable
in the ebuild as opposed to "HEAD".  Also, the SLOT variable should be put back
to "22.0.91" so that the ebuild puts the application into the correct slot. 
I'll post a patch shortly to the v1.2 ebuild that Christian posted a little
while ago.

The way the ebuild is currently setup, the 22.0.91 ebuild is building the
22.0.92 version of emacs (!), or, more correctly, what it is actually building
is the head revision of the repository, which is *currently* 22.0.92.  It'll
change again soon and this ebuild will break again unless something is done.

Other ebuilds are also in error:
* emacs-cvs-22.0.92.ebuild, ECVS_BRANCH should be "EMACS_PRETEST_22_0_91".
* emacs-cvs-22.0.50-r{1,2,3}.ebuild should be removed from Portage because the
tags no longer exist in the CVS repository.

George, you may want to re-open this bug to prevent unnecessary duplication. 
It appears that a 3rd party can't.

Best regards,
John

Ack, typos:

Other ebuilds are also in error:
* emacs-cvs-22.0.90.ebuild, ECVS_BRANCH should be "EMACS_PRETEST_22_0_90".

Sorry.

- John

Created an attachment (id=105015) [details]
Patch to correct the CVS tag the ebuild fetches.

Created an attachment (id=105016) [details]
Similar patch for the emacs-cvs-22.0.90 ebuild.

Created an attachment (id=105018) [details]
New ebuild for emacs 22.0.92 (current CVS HEAD revision), just for good
measure.

I will wrangle this issue when I am back home and can test a bit better (this
notebook isn't suited)...On 02 Dec 07 I will have some more time maybe, latest
03 Dec.

I haven't forgotten you, just was busy, because my mail program crashed badly,
which gave me an excuse to finally switch.  22.0.92 is in the tree now, please
test, 22.0.91 has been corrected, all older versions have been punted.  As it
won't build on my two system I can't really test here.

(In reply to comment #8)
> Created an attachment (id=105018) [edit] [details]
> New ebuild for emacs 22.0.92 (current CVS HEAD revision), just for good
> measure.
> 

I tried to update my emacs-cvs and yes ebuild was successful but here are some
observations:

When I did emerge --pretend emacs-cvs I got emacs-cvs-22-.0.91 but the built
application shows emacs.emacs-22.0.92 and emacs-22.0.92. So i think something
is still wrong.

I also checked the version (emacs --version) and it says it is 22.0.92.1. I
don't now if this is expected or not. 

Please let me know if this is not what we expect. I don't know if I need to
re-open this issue. 

Note: I did emerge-webrsync last 03Jan2007.

(In reply to comment #11)
You'll need to sync again.  The fixes were just posted a few hours ago and thus
your emerge-webrsync on the 3rd would have not gotten them.

Christian, I'm compiling against the new ebuilds right now on two different
machines.

- John

(In reply to comment #10)
> I haven't forgotten you, just was busy, because my mail program crashed badly,
> which gave me an excuse to finally switch.  22.0.92 is in the tree now, please
> test, 22.0.91 has been corrected, all older versions have been punted.  As it
> won't build on my two system I can't really test here.
> 
Christian,

I've built 22.0.91 and 22.0.92 successfully on an AMD K7 and an old Pentium 1. 
It works fine and it builds what it says it builds.  Thanks so much for the
help.

One thing surprised me but I've determined that it is intended behavior.  The
ebuild uses the alternatives_auto_makesym() function in the alternatives eclass
to set symlinks for slotted packages to the latest version installed regardless
of whether or not you're currently emerging the latest version.  It was a
little bit of a surprise to me while emerging 22.0.91 with 22.0.92 already
installed to have the symlinks pointing to the 22.0.92 version at the end of
the build.  However, like I said, it's intended behavior.

The ebuild is *very* well behaved, though.  I unmerged the 22.0.92 version and
the symlinks were automagically placed on the 22.0.91 version.  Cool!

- John

(In reply to comment #13)
> (In reply to comment #10)
> > I haven't forgotten you, just was busy, because my mail program crashed badly,
> > which gave me an excuse to finally switch.  22.0.92 is in the tree now, please
> > test, 22.0.91 has been corrected, all older versions have been punted.  As it
> > won't build on my two system I can't really test here.
> > 
> Christian,
> I've built 22.0.91 and 22.0.92 successfully on an AMD K7 and an old Pentium 1. 
> It works fine and it builds what it says it builds.  Thanks so much for the
> help.

 That's our job here...but you also provided good help here, thanks.

> The ebuild is *very* well behaved, though.  I unmerged the 22.0.92 version and
> the symlinks were automagically placed on the 22.0.91 version.  Cool!

 Thank Matthew Kennedy here, he made up most parts of the ebuild, I think.