[Aptitude-devel] Bug#745425: aptitude: dependency handling jammed on chromium upgrade

[Edward Welbourne]

> For example, when you saw chromium 34.0.1847.116-1~deb7u1 and marked
> for installation (an upgrade of the version of the browser, but
> changing the package to that targetted to the stable distribution),
> chromium-inspector should have been marked to change to the same
> version targetted for stable, 34.0.1847.116-1~deb7u1 (I believe that
> they always require to upgrade in lock-step).  Or the same with
> 34.0.1847.116-1 (without "~deb7u1"), if it was available in your
> mirrors.

except that, when I positioned the cursor on the 34.*~deb7u1 version
that was listed and typed +, aptitude selected the 33.* version rather
than the 34.* version, presumably because it had reasons to prefer a
testing version over a stable one.  It did not report any other 34.*
versions.

> As I said above, only the versions of chromium* are not enough, there
> are many other things at play behind the scenes.

Fair enough.  Unfortunately, I now only have the information I reported,
having long since updated plenty more things.

> I think, as I said above, that in that case telling to aptitude to
> keep v 33 of both and everything would have been fine, bugs aside.
> (So, same result of what happened, just without reinstall).

except that doing this involved putting inspector on hold when it
claimed to have a security fix to which it wanted to upgrade.

> If you don't know about the feature, you can press 'e' when there are
> conflicts, and then '.' and ',' to navigate forward and backward the
> solutions offered, and press '!' if one of them satisfies you.

I did know.  I imagine I looked through the options and found the least
bad was to hold inspector at its present version, so settled for that,
despite having reason to believe there was a major security issue in it.

> You can set priorities to the repositories, see "man apt_preferences",

Thanks for the tip ;-)

> In general, these things are tricky and testing is not always very
> stable ;)

... which is, after all, the point - and why we report bugs when we find
them, so that they can be fixed before they reach stable ...

It sounds like we don't have enough information to pursue this issue
further; and I haven't seen similar symptoms in a while.  So there are
probably more productive uses for your time than investigating further,

	Eddy.