PALUDIS_UNPACK_UNRECOGNISED_IS_FATAL is being set for 0/1 EAPIs

[bobgoat]

I screwed something up and decided to reemerge world to be safe. Last time I did this was with Paludis 0.26.something I believe and there weren't any problems. With 0.28.0 though it fails on eselect-news:

# paludis -i1 eselect-news

>>> Unpacking news.eselect-20080320 to /var/tmp/paludis/app-admin-eselect-news-20080320/work
unpack: error: Don't know how to unpack /usr/portage/distfiles/news.eselect-20080320
unpack: making ebuild PID 5605 exit with error
die trap: exiting with error.

I re-installed tar, bzip and paludis to be on the safe side but to no avail. If you need any more info I will gladly post it but other than using Paludis my system is pretty typical. x86, using the latest non-masked version of all the system stuff including gentoo-sources kernel. The only USE flag I have enabled for paludis itself is python.

[ciaranm]

paludis --info?

[shade]

paludis -i freeciv --log-level debug

[shade]

paludis --info

[shade]

/usr/share/paludis/eapis/0.conf

[ciaranm]

Ok, someone who can reproduce this who knows gdb needs to work out why PALUDIS_UNPACK_UNRECOGNISED_IS_FATAL is being set to yes.

[bobgoat]

paludis --info

[bobgoat]

OK, I've attached the output of paludis --info.

[ciaranm]

Someone who can reproduce the problem, please try noisy-log-messages.patch. It won't fix it, but it might tell us why the environment variable is being missset.

[quidome]

noisy log output

[ciaranm]

So this is where it's going wrong:

  ... We just did EAPI '0'
paludis@1216651952: [WARNING GREPFORTHIS] (same context) k.get("unpack_unrecognised_is_fatal") is 'false'
paludis@1216651952: [WARNING GREPFORTHIS] (same context) destringify<bool>(k.get("unpack_unrecognised_is_fatal")) is 'false'
paludis@1216651952: [WARNING GREPFORTHIS] (same context) unpack_unrecognised_is_fatal is 'true'

The question is why...

[ciaranm]

For anyone who's seeing this and who wants a quick fix, add this to /etc/paludis/bashrc:

PALUDIS_UNPACK_UNRECOGNISED_IS_FATAL=""

AND REMEMBER TO REMOVE IT ONCE WE GET THIS FIXED'''

But this is not a fix, and I still need someone who can reproduce this to use gdb to tell me why the assignment in the previous comment is going wrong.

[ciaranm]

Someone who can reproduce the problem:

Please apply the two patches I just attached to 0.28.0, and confirm that this fixes things.

I'll do a 0.28.1 if this works.

[bobgoat]

I can confirm that those patches have fixed it for me.

[ciaranm]

0.28.1, thanks.