Reordering in package changelogs (was Re: rawhide report: 20100129 changes)

Panu Matilainen pmatilai at laiskiainen.org
Tue Feb 2 12:52:35 UTC 2010


On Tue, 2 Feb 2010, Nicolas Mailhot wrote:
>
> Le Mar 2 février 2010 11:35, Kevin Kofler a écrit :
>>
>> Nicolas Mailhot wrote:
>>> That is your interpretation. I see nothing on this page to support this
>>> claim. And actually it is contrary to format #3 logic, since its main
>>> difference with other changelog formats is that the version is not part of
>>> the entry header, so there's no reason to limit one entry to one version
>>
>> It's blatantly obvious that all these formats have one thing in common:
>> there's one entry per new EVR.
>
> Again, this is your interpretation
>
>> Our automated tools (e.g. make clog as
>> already pointed out by Michael Schwendt) also expect that.
>
> IIRC when this guideline was written rpm devs complained the EVR was never
> supposed to be in the changelog header, it was supposed to end at the email
> mode and the EVR was an emacs rpm mode addition. Nervertheless FPC chose to
> accept existing practices.

This still applies: in the changelog styles #1 and #2, the EVR ends up 
being a part of the email address which is hardly "right".
     rpm -q --qf "[%{changelogname}\n]" <pkg>

> This changelog style conforms to the existing spec, it has been in use in
> Fedora for several years, it may surprise you, but changing the spec
> retroactively is not the way to prove your point.

The problem in the example you provided in 
http://lists.fedoraproject.org/pipermail/devel/2010-February/130062.html 
is that it lumps two EVR's into a single changelog entry. That just 
doesn't make sense.

Taking the same changelog snippet (style #1):
* Thu Jan 28 2010 Michael Schwendt <mschwendt at fedoraproject.org> - 2.2-10
- Fix tuple_copy() further (it was completely broken as the mowgli
   dict wasn't copied at all).

* Thu Jan 28 2010 Michael Schwendt <mschwendt at fedoraproject.org> - 2.2-9
- Let set_tuple_cb() work on a copied tuple
   (the metadata updates flood is too racy IMO).
- Fix tuple_copy().

In style #3 it would be:
* Thu Jan 28 2010 Michael Schwendt <mschwendt at fedoraproject.org>
- 2.2-10
- Fix tuple_copy() further (it was completely broken as the mowgli
   dict wasn't copied at all).

* Thu Jan 28 2010 Michael Schwendt <mschwendt at fedoraproject.org>
- 2.2-9
- Let set_tuple_cb() work on a copied tuple
   (the metadata updates flood is too racy IMO).
- Fix tuple_copy().

 	- Panu -


More information about the devel mailing list