Mailinglist Archive: opensuse-packaging (206 mails)

[Cristian Morales Vega <cmorve69@xxxxxxxx>]

2011/1/19 Richard Guenther <rguenther@xxxxxxx>:
> On Tue, 18 Jan 2011, Cristian Morales Vega wrote:
> > The second one IMHO is an error in the rpmlint check. But I don't know
> > the history, the "broken" part perhaps fixes something else. The thing
> > is that if all the libraries in a package show the
> > "shlib-unversioned-lib" warning then the "shlib-policy-missing-lib"
> > will also be shown. Since you have a single library and is detected as
> > unversioned...
>
> The check is supposed to catch the .so symlink that belongs into
> a -devel package (and doesn't count as a shlib) vs. a (versioned) shared
> library which always has a version after .so.  In fact an unversioned
> library is both the devel and the shlib part.  One could restrict
> the check so that non-symlink .so files are allowed, but unfortunately
> that would bogously trigger for linker-script .so files as well, so
> it will be not that simple to avoid the warning for your case
> (and it is indeed useful in general, so dropping it isn't an
> option either).

I continue without knowing Python ;-) But the part that adds the
libraries, "libs.add(bi.soname)", is inside an "if
stat.S_ISREG(files[f].mode):"
But indeed, linker-scripts are a problem. Anyway it's just a very
uncommon case vs a really very uncommon case, I would not touch
rpmlint because of this.
--
To unsubscribe, e-mail: opensuse-packaging+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-packaging+help@xxxxxxxxxxxx