Bug#717995: [ats@offog.org: Re: RFS: feedparser/5.1.2-2]
----- Forwarded message from Adam Sampson <ats@offog.org> -----
Date: Mon, 29 Jul 2013 17:13:42 +0100
From: Adam Sampson <ats@offog.org>
To: Etienne Millon <etienne.millon@gmail.com>
Subject: Re: RFS: feedparser/5.1.2-2
Hi Etienne,
Thanks very much for looking at the package -- much appreciated!
Do you mind if I attach a copy of this to the ITA bug?
On Mon, Jul 29, 2013 at 02:45:26PM +0200, Etienne Millon wrote:
> Incidentally, getting rid of the embedded feedparser copy was the
> first thing I did with rss2email too :)
It took quite a bit of work in rawdog since I used to have a load of
local patches to feedparser -- it's only really become possible with the
more predictable sanitisation that recent feedparser does.
> I think that you should say what you changed, or just put "(no changes
> needed)" in case you went through the upgrading checklist and found no
> problems.
Done:
* Check that the package meets Debian policy version 3.9.4 (no further
changes needed), and update Standards-Version.
> The python dependencies may actually stay Build-Depends-Indep, but I
> don't think that the distinction is actually used. So, do as you wish.
I'd moved them to Build-Depends because that's what the Python policy
and LibraryStyleGuide use in their examples, but reading the rationale
it sounds like Build-Depends-Indep is appropriate, so I'll go with that.
> I believe that it's missing the copyright info relative to debian/*
> files.
Good catch -- I was intending it to be covered by the * section, but of
course that's missing Decklin's copyright. Fixed.
> > diff -Nru rawdog-2.13.dfsg.1/PKG-INFO rawdog-2.16/PKG-INFO
> > --- rawdog-2.13.dfsg.1/PKG-INFO 2010-10-15 23:38:53.000000000 +0200
> > +++ rawdog-2.16/PKG-INFO 2013-07-15 23:45:30.000000000 +0200
> > -License: GNU GPL v2 or later
> > +License: UNKNOWN
>
> I don't think that this is wanted :) That's probably because you
> removed the "license=" key in setup.py.
Yes, I'd spotted that -- I think it's a distutils oddity. The distutils
manual says:
"The license field is a text indicating the license covering the
package where the license is not a selection from the “License” Trove
classifiers."
... so I'd removed it when I added the license to the classifiers list.
I'm not too worried since PKG-INFO and the PyPI page do correctly
include the license classifier.
(The same goes for "Platform: UNKNOWN", but rawdog never had a platform=
kwarg anyway -- and in that case the distutils documentation doesn't
even say what values are permitted...)
> Great! I didn't know about feedfinder and the related license issue.
> Thanks for maintaining feedscanner too. I might push support for this
> useful library to upstream rss2email. Would you be interested in
> packaging it separately? (not necessarily now of course).
I was hoping you might ask that ;-) Yes, the intention is to spin it off
as a standalone library after it's been shaken down a bit in rawdog; it
has the same (trivial) API as feedfinder, so it should be a drop-in
replacement.
> > <urlopen error [Errno 2] Aucun fichier ou dossier de ce type:
> > 'missing.rss'>
That's one downside of writing software in the UK! I've made test-rawdog
set LC_ALL=C.
> > Test failed: expected testauto/output.html to contain
> > 'HEADING-01-01-18:00'
... and that's another one! 18:00 on January 1st is 18:00 local time for
me, but not in other timezones. test-rawdog now sets TZ=UTC.
I've also added testcases to make sure that varying locale and timezone
does actually have the desired effect on date formatting. (And spotted
another encoding problem that I'll need to fix...)
Thanks very much,
--
Adam Sampson <ats@offog.org> <http://offog.org/>
----- End forwarded message -----
Reply to: