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

Re: How to fix package-has-a-duplicate-relation?


[ please do not send me a identical mail besides that mail you sent to
-mentors. I _am_ subscribed to -mentors. Thanks! ]

Thomas Viehmann wrote:
> Rene Engelhard wrote:
> > Sven Luther wrote:
> >>Or you could try a shlibs override. look at dpkg-shlibsdep manpage for
> >>more details.
> > Or you just kick the first xlibs out using sed :)
> Given the fact that
> - There is no policy reference in the Lintian output,
>   (Though I may have overlooked the item in policy to specify this.)
> - The explanation "can break some tools" is rather soft...
> - ... especially when there are approximately 307 other packages that would
>   break the same tools,
> - The bug is not in the package and is "self-fixing" as far as the package
>   itself is concerned,

When dpkg-dev is fixed or when the xlibs split occurs. Yes.

> - The bug would have been fixed in dpkg-shlibdeps and/or xlibs if anyone cared
>   enough,
> it may be argued that all you're doing is a "my package has zero or less lintian
> errors" beauty contest, at least partially with *dubious* methods:


> Both of your suggestions cause unneccessary problems when backporting packages
> which use it. Especially with the sed method, you're not removing an incorrect
> dependency: You unconditionally remove a depencency, which is a bug in itself,

Oh, yes. woody & sarge users would have problems with backporting....

> unless, of course, you specify a build-dependency on xlibs (>> 4.2.0). (And no,
> I'm not suggesting the latter.)


> That type of hack may be justified when the problem you cure is more grave than
> the side effects, but given that people are porting packages.

What are the problems with porting here? Can you explain me?
Porting isn't a problem. Backporting maybe.

> > - who wonders why people don't read the archives; this issue came up already
> > a few times...
> For the individual that's because (# of issues * probability that I forget to
> search or overlook something)>0 and for the issue (# of people * probability

That was more or less a intuitive subject so if you have searched you
would have found it...

> that a random person does the same thing)>0.


Anyway, I'll remove this sed with the next upload...
(kover 2.9.1-2)



Attachment: pgpemZgWvctdI.pgp
Description: PGP signature

Reply to: