People that install from stage 2 or 3 have several libtool archives on their system with hardcoded paths that are dependent on the setting of CHOST used to build their stageball. If they later change CHOST (often from i586 to i686), and update gcc, their systems will encounter strange and very broad compilation failures. http://forums.gentoo.org/viewtopic.php?t=65846 describes one way to get around this issue afterwards, but it is very crude and could hurt things like cross-compilers. I have spoken to Azarah about this issue, and he will consider ways of extending fixlafiles.awk to address it, but in the meantime, we both agree that it would be a good idea to warn people that do stage 2 or 3 installs to stay with the CHOST that their stageball was compiled with.
I'll take it
I've committed the changes to the document. I won't resolve this bug though so it keeps reminding me to regularly check if the problem is still valid.
committed. please keep me informed about changes!
Seems like this stays until the end of times. Fixing.
The problem still exist: - italian topic: http://forums.gentoo.org/viewtopic.php?t=74473), - a lot of same issues seen on irc Is it sufficient to force re-emerge (from source, not binaries) libtool & gcc?
I can't read Italian, but you may *not* change CHOST if you're not using stage1. This bug is set to "FIXED" because the warning not to change CHOST is implemented.
mmmhhh... the CHOST used to compile stages is i586? And: anyway, is sufficient re-emerging libtool & gcc to avoid problems after changing CHOST? The post I linked solved the problem such this way.
I guess Robert is the best person to reply on this. rac?
