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

Re: RFC: Adding discussion about required versions to Python policy



On Monday, March 19, 2012 12:20:18 PM Jakub Wilk wrote:
> * Scott Kitterman <debian@kitterman.com>, 2012-03-17, 14:29:
> >Upon reflection, this could be better stated something like this:
> >
> >"The generated minumum dependency may be different than the lowest
> >version currently supported. In such cases, X-Python-Version must
> >still be specified if the generated dependency is not sufficient."
> >
> >To give a specific example, even though python3.2 is the only supported
> >python3, for an arch all module, dh_python3 will generate a dependency
> >of python3 >= 3.1.3-13.  If the upstream code requires 3.2, then you
> >still need to specify (in this example) X-Python3-Version.
> 
> This is true, but I'm not sure why Python Policy needs to talk about
> this. If it does, then probably appendix B would be the correct place.
> Or a footnote.
> 
> In general, how X(S)P(3)V is translated to dependency on python(3)
> varies depending on which helper you use. A packaging helper can add a
> dependency on "python(3) (>= $V)" for several reasons:
> 
> 1) because the package declared "X(S)P(3)V: >= $V" [all helpers];
> 
> 2) because the package ships extension modules (or other files that
> cannot be shared across versions) only for versions >= $V [all helpers];
> 
> 3) because the helper is implemented in such a way that it supports at
> runtime only (a subset of) versions available at buildtime [dh_python2,
> and sometimes python-central];
> 
> 4) because it generates maintainer scripts that need such version of
> python(3) [dh_python2, dh_python3].

I think something like my first paragraph needs to go wherever it says X3-P-V 
can be omitted if all versions are supported, because it has to be clear that 
"all" isn't just all in Unstable on the day you're uploading.  It's a function 
of how the dependencies are calculated.  

I think your helper specific information ought to go in appendix B.  

Does that make sense (to the extent anything less than a rewrite makes sense)?

Scott K

Attachment: signature.asc
Description: This is a digitally signed message part.


Reply to: