On Mon, Jun 20, 2005 at 07:06:49PM +0200, Filippo Giunchedi wrote: > [ resending to BTS, forgot to address b.d.o] > > On Sun, Jan 02, 2005 at 05:50:30PM +0100, Adeodato Simó wrote: > > Hi, > > > > under some circumstances, it can be nice for dch to read the > > Maintainer field from debian/control and create the new changelog > > entry from there. > > actually debchange should set email/fullname to the last entry in changelog if > getpwuid/email-guesses fails. this is however extremely rare (if not > impossible > at all). > would an option to ignore env and/or various guesses and set email/fullname to > the previous changelog entry be ok?
I don't think so, I really think too that dch should provide a way to
use the debian/control Maintainer field. It already detects when a
package is co-maintained and adds fancy
[ $DEBNAME ]
* ...
* ...
but does the work half, since it uses $DEBEMAIL to generate the
email/fullname of the Changelog entry. debian/control Maintainer field
is mandatory, we should use it, not the last changelog entry.
In fact, I think dato's patch is not the right way to do it. there
should not be any -c option : we have 2 cases :
* the package is nmu-ed and the nmu-er has to use dch -n, and his name
will be used using the env or any other method.
* the package is not nmu-ed, then dch has to :
- use debian/control Maintainer field for the full name `signing'
the changelog entry
- use DEBEMAIL and alike in the env to generate the fancy output I
was talking about sooner.
this is the logical behaviour for all the situations I can think of.
and if one wants to keep the current behaviour, then maybe dch should
provide a --no-control meaning that it should not use debian/control to
generate the changelog.
--
·O· Pierre Habouzit
··O [EMAIL PROTECTED]
OOO http://www.madism.org
signature.asc
Description: Digital signature
