Re: Validation failure for libpng15-32bit

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 2/4/12 6:26 AM, Sjors Gielen wrote:
> Op 03-02-12 03:13, Alexander Hansen schreef:
>> What?  Are you proposing to have packages that link to both
>> libpng15 and libpng15-32bit?
>> 
>> If not, our standard practice is to have the BuildDependsOnly
>> parts of related packages Conflict/Replace, even when the two
>> package families have no file overlap.  It avoids accidentally
>> building against the wrong one.
> 
> The libpng15-32bit library does not install any headers; it uses
> those of libpng15, which are the same ones. I can change this to a
> Conflicts, and adding the same headers again, if you think that's
> better. (The only thing where the two might possibly lead to a
> build error at the moment is by having both shlibs installed at the
> same time, which can't be uninstalled per Fink policy since they
> might have dependencies.)
> 

Ah, I missed the lack of headers.

I'd probably still argue for expanding libpng15-32bit to include
headers, and carry a Conflicts/Replaces libpng15.  Not having any of
libpng15 installed when libpng-32bit is, and vice versa, seems to me
like it makes the separation between the 32-bit and 64-bit packages
more robust.

>> You *didn't* mention that it FAILS VALIDATION!
>> 
>> Looking over a dpkg -L listing is well and good, but come on
>> now.
> 
> I'm wondering: is your problem with this situation that the
> package installs into %p/lib32 or is the problem that it fails
> validation?

The latter.  %p/lib32 is a counterpart to the existing %p/lib64 for
64-bit packages on 32-bit.

 I
> knew it failed validation, but considered it to be OK since I
> explicitly mentioned lib32 in response to a question from you once
> (don't remember if it was IRC or a mailing list, but we talked
> about this), and IIRC you seemed to agree with the solution (even
> though I probably didn't mention what the validator thought).
> However, if %p/lib32 is OK by your standards, but failing
> validation is not, what meaning does the validation failure still
> have?

I'm not the project, but one person. :-)

My recollection is that I mentioned at that time that we would need to
extend fink to handle %p/lib32.

> 
> Either way: would you consider the best solution in this case be to
> a) patch the validator, b) change the installation directory, or c)
> offer my sincere apologies for not fully informing you beforehand?
> 
> Sjors

a), if nobody objects to the addition.  I'd call it an "enhancement"
rather than a "patch".  Doing just this part is trivial.

I'd even go further and suggest that it would be worth adding code to
provide a more automatic handler for 32-bit builds on x86_64, as a
counterpart to the 64-bit option for i386 Finks on machines that can
handle it.  That's a bit more involved than just tweaking the
validator's allowed paths, but it provides a more convenient way to
deal with similar packages in the future.

I tend to get quite excitable about packages that don't pass .deb
validation, because I had a Fink openssl release disable my entire
system in the pre-validator days by wiping my /var directory.  This
isn't that bad. :-)
- -- 
Alexander Hansen, Ph.D.
Fink User Liaison
http://finkakh.wordpress.com/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk8tf9EACgkQB8UpO3rKjQ81agCfYG1RDWjaPXJyEwb/qkNs/Kco
FScAn26pEWR2e1D5jXS3/UvmxvowJiME
=qIYS
-----END PGP SIGNATURE-----

------------------------------------------------------------------------------
Try before you buy = See our experts in action!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-dev2
_______________________________________________
fink-core mailing list
[email protected]
List archive:
http://news.gmane.org/gmane.os.apple.fink.core
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-core