Jump to content
funtoo forums
Sign in to follow this  
Oleg Vinichenko

pam updates

Recommended Posts

Hi there!

Everyone, recently in Gentoo, multilib changes made to pam ebuild, which results in pam ebuild downgrade. Funtoo has this package forked and it's not yet ported to newer multilib. Before that, it is strongly recommended to NOT let portage downgrade this package, otherwise system may stuck, freeze on loging or uncertain, permission related issues observed. TO avoid pam downgrade, during usual auDN arguments add -exclude argument, so it will look like this

 

# emerge -auDN @world --exclude pam.

 

This will be reported on bugs.funtoo.org and fixed.

 

Share this post


Link to post
Share on other sites

Oleg, I think I am going to need to create a script to track these multilib changes. They appear to be an effort to get rid of the 32-bit "emul" packages in Gentoo. I am not certain yet as to whether this is the right approach. It certainly makes the ebuilds a mess.

Share this post


Link to post
Share on other sites

I don't know about the long-term plans, but the new multilib approach is much more elegant (at least in theory, maybe not at ebuild level). Instead of depending on a bunch of pre-compiled emul-linux- ebuilds (which contains specific versions of a set of packages, and may be hard to maintain), multilib ebuilds can depend only on what they need, and the libraries will be built from from source for every needed architecture.

 

Now, the transition to this may be hard. A lot of ebuilds depend on emul-linux- ebuilds, and these now start depending on native multilib ebuilds (instead of providing the pre-compiled binaries). I think this is temporary (as new ebuilds should depend on new multilib ebuilds instead of emul-linux- ones), but it might be very long to migrate everything.

Share this post


Link to post
Share on other sites

Jean-Francis, I tend to agree. The purpose of me asking (see here) is to figure out how important this is. It does appear to still be important for desktops. But not for servers.

 

The problem we will face is for forked ebuilds, like in this pam example. If gentoo adds the multilib support to the ebuild, and we have previously forked it, and then starts depending on it in other ebuilds, then it will create an unresolved dependency.

 

I think that the best solution is to have an automated report -- I am figuring out the best place to put this -- that will track ebuilds in gentoo with multilib support, and alert us if any of our forked ebuilds need to be updated. I have a beta script to do this. I am trying to figure out the best way to integrate it. It may be during Portage tree regeneration. I can generate a JSON report on ports.funtoo.org and then have it display on the wiki. This would be similar to our Compare Forked Packages to Gentoo report.

 

I think we will be needing more pro-active QA checks like this in the future...

Share this post


Link to post
Share on other sites
Sign in to follow this  

×
×
  • Create New...