|
Re: rpm naming convention libfoo vs libfoo-devel
David Mathog wrote:
> Is there supposed to be a naming convention such that
>
> libfoo
>
> has a corresponding
>
> libfoo-devel
>
> ?
>
> If so, there are some violations in 2008.1, for instance, consider
> these:
>
> libfftw-devel-3.1.2-8mdv2008.1
> libfftw3-3.1.2-8mdv2008.1
>
> Originally my system had libfftw3, so I tried:
>
> % urpmi libfftw3-devel
>
> and that failed, because the name was not as expected.
>
> I looked into this a bit further and it seems that some libraries are
> like this one:
>
> libxvmc1-1.0.4-2mdv2008.1
> libxvmc1-devel-1.0.4-2mdv2008.1
The above should be the libraries for use, and the libraries
for development, of the same libraries. If you are not compiling,
you don't need the devel package.
>
> and others like this one:
>
> libxxf86dga1-1.0.2-2mdv2008.1
> libxxf86dga-devel-1.0.2-2mdv2008.1
This should be libraries for use for one set of libraries and
the libraries for development of a different set of libraries
(dga1 and dga being different sets of libraries).
>
> Does it really make sense to do it both ways???
If the second example pertained to the same libraries, someone
made a typo.
>
> Also, what's the point of duplicating some of the version number onto
> the library name? There are certainly a few packages which have a
> numeric identifier unrelated to the version number (libstdc++5 vs.
> libstdc++6) but for the vast majority of libraries "libfoo" and a
> version number "x.y.z" are sufficient to identify the rpm.
Different sets of libraries. The basic functionality may be the
same, but they are different, and normally are not interchangable.
jim b.
--
UNIX is not user unfriendly; it merely
expects users to be computer-friendly.
|