[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFS: poco and poco-doc (updated packages) [3rd try]



On Sunday 08 June 2008, Russ Allbery wrote:
> George Danchev <danchev@spnet.net> writes:
> > On Saturday 07 June 2008, Krzysztof Burghardt wrote:
> >> This looks reasonable, but trigger another lintian warrning:
> >>
> >> N: Processing binary package libpocoxml5-dbg (version 1.3.2+dfsg1-1) ...
> >> W: libpocoxml5-dbg: package-name-doesnt-match-sonames libPocoXMLd5
> >
> > sure, lintian has provided you with oneliner (as copied verbatim from
> > libpkg-guide #3 Naming shared library packages) to determine the correct
> > package name out of the soname, but unfortunately it doesn't seem to be
> > working for object files containing debugging information (I didn't
> > check why, will probably do;-):
> >
> > $ objdump -p usr/lib/debug/usr/lib/libPocoXMLd.so.5 |
> > sed -n -e's/^[[:space:]]*SONAME[[:space:]]*//p' |
> > sed -e's/\([0-9]\)\.so\./\1-/; s/\.so\.//'
>
> I think there's something more fundamentally wrong here.  If this is a
> regular shared library, not detached debugging symbols, it's in the wrong
> directory.  The *only* thing that should be in /usr/lib/debug is detached
> debugging symbols.  

Yes, dh_strip -k was called to split debigging symbols in a separate file 
(containing the detached debugging symbols) in usr/lib/debug/, in order to 
avoid binary duplications with things we want debugable, but the above 
oneliner produces no package name for so generated object file. 
Turning back to dh_strip --dbg-package=... the above oneliner (as also used 
internally by lintian I believe ?) produces a package name:

objdump -p debian/libpocoxml5-dbg/usr/lib/libPocoXMLd.so.5 | 
sed -n -e's/^[[:space:]]*SONAME[[:space:]]*//p' | 
sed -e's/\([0-9]\)\.so\./\1-/; s/\.so\.//'
libPocoXMLd5

shared library goes in /usr/lib and as expected lintian complains with:
libpocoxml5-dbg: package-name-doesnt-match-sonames libPocoXMLd5
because of the missing 'd' before '5', at least, hence that leads us to a 
package name as `libpocoxmld5-dbg', is that correct ?

> (There's a lintian check for that as well; I'm not 
> sure why it's not triggering -- maybe I'm misunderstanding what's going
> on?)

the goal was/is to generated -dbg packages that are using separate (detached) 
debugging info and stored in /usr/lib/debug/, thus do you reference to check 
described in #299578 ?

> If you're shipping a debugging version of the shared library that's a full
> shared library in its own right because building with debugging changes
> the library, then yes, you'll need to override a warning about the package
> name.  But it should be in /usr/lib.

Just to make it crystal clear for myself, a package name override is needed 
because of the ending -dbg only, right ?

-- 
pub 4096R/0E4BD0AB 2003-03-18 <people.fccf.net/danchev/key pgp.mit.edu>
fingerprint 1AE7 7C66 0A26 5BFF DF22 5D55 1C57 0C89 0E4B D0AB 


Reply to: