Re: Would lcms really break 309 packages?

[ilmari@ilmari.org (Dagfinn Ilmari Mannsåker)]

Nathanael Nerode <neroden@twcny.rr.com> writes:

>  From bjorn.haxx.se:
>
> 20 packages wait for lcms (63 days old, Valid candidate, breaks 309 pkgs)
>
> Really?  This seems unlikely.

Most of those seem to be indirect dependencies and don't really count.
The problem is that testing has liblcms, while unstable has liblcms1,
and stuff like Qt (via libmng1) and imagemagick depend on it, causing
massive recursive "breakage".

The only packages in testing that depend directly on liblcms are
autotrace, imagemagick, libmng, prestimel, rubymagick and scribus.

Of these, imagemagick also has different sonames in testing and
unstable, so all its direct dependencies would also be affected, these
are: autotrace, gnuift, php3, prestimel and rubymagick.

> Perhaps it needs hinting?

I think that script tries a bit too hard to find breakage and thus turns
up a bunch of false positives. I believe (and hope) the scripts that
actually make the decisions are more accurate in assessing breakage.

-- 
ilmari