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

Re: Upcoming QOF transition



On Wed, 6 May 2009 12:25:04 +0200
Adeodato Simó <dato@net.com.org.es> wrote:

Final plan:

1. I've put QOF 0.8.0 into string freeze today and asked for translation
updates - deadline in 10 days time.

2. Once QOF is out of string freeze (10 days, 18th May), I'll release
QOF 0.8.0 upstream and upload 0.8.0-1 to Debian to join the NEW queue.
libqof2 0.8.0-1 will contain the qof-data package for Policy 8.2.
(qof-backend-gda needs more work upstream and won't be part of 0.8.0)

3. Once QOF 0.8.0 clears NEW, I'll upload new upstream versions of
gpe-expenses and pilot-qof that depend on the virtual backend package(s)
instead of the actual backend names to make pilot-qof and
gpe-expenses binNMU safe. Also fix the issue with libqofexpensesobjects0
in the same gpe-expenses upload by adding a new package,
libqofexpensesobjects-data. pilot-qof needs to go through NEW
because that upstream release adds new packages too. 

4. Goedson, if you don't have time to upload a fixed gnotime before QOF
comes out of NEW, I'll prepare an upload containing the fix I sent to
you (and which IIRC you've also sent upstream) as a new Debian version
using the existing dpatch support in gnotime, to prevent a FTBFS bug
against gnotime. The fix is safe for libqof1 0.7.5-1 currently in Sid,
so it doesn't need to wait for libqof2 if you prefer - a binNMU could be
done later. (gnotime will be the first binary package using libqof2
because it won't need time in NEW.)

5. During the time pilot-qof and gpe-expenses are in NEW, the current
versions will be buildable locally with libqof2 (subject to the gcc
default version staying at 4.3 as the gpe-expenses upload fixes an
issue with gcc-4.4.) The issue will be the change in the backend
package names as the current packages don't know about the new
Provides:. It'll only be during the time in NEW so there is probably no
point putting pilot-qof and gpe-expenses into NEW until QOF clears NEW
or we'll end up with needing some binNMU's at the end.

This saves one complete run through NEW and it removes the need for
bin-NMU's of pilot-qof or gpe-expenses, possibly gnotime too. Thanks for
the suggestion, Adeodato.

-- 


Neil Williams
=============
http://www.data-freedom.org/
http://www.nosoftwarepatents.com/
http://www.linux.codehelp.co.uk/

Attachment: pgpgSvnWxJAcZ.pgp
Description: PGP signature


Reply to: