Re: Source for KOffice 1.6.2 has been uploaded

On Thursday, 15. February 2007 11:51, Jarosław Staniek
wrote:

> > You (Jaroslaw) cannot possibly be serious, both
about libpqxx's decision
> > to retire a long-obsolete API being
"ugly" and suggesting that packagers
> > should ignore an important update just so kexi
will compile.
>
> The changes in pqxx we have problem with are related to
API. I think we
> agreed this is not something one expect from minor
release.

The lesson to be learned there is to improve communication
with upstream 
client library developers, I suppose. In any case, laying
blame is beside the 
point, the damage is done.

> Especially that 
> the topic is covered when KOffice is tagged, so patches
that "at least
> compile" are not enough, although thanks for
providing them - we can
> publish them within the release notes for users of pqxx
2.6.9.

How about testing them then? Tags can be slipped, releases
can be delayed. 
There's not much point in releasing bugfix releases with
known bugs.

-- 
   ,_,   | Michael Nottebrock               | lofifreebsd.org
 (/^ ^) | FreeBSD - The Power to Serve     | http://www.freebsd.org
   u/   | K Desktop Environment on FreeBSD | http://freebsd.kde.org

_______________________________________________
Kexi-devel mailing list
Kexi-develkde.org
http
s://mail.kde.org/mailman/listinfo/kexi-devel

Michael Nottebrock said the following, On 2007-02-15 12:53:

 >>> You (Jaroslaw) cannot possibly be serious,
both about libpqxx's decision
 >>> to retire a long-obsolete API being
"ugly" and suggesting that packagers
 >>> should ignore an important update just so kexi
will compile.
 >> The changes in pqxx we have problem with are
related to API. I think we
 >> agreed this is not something one expect from minor
release.
 >
 > The lesson to be learned there is to improve
communication with upstream
 > client library developers, I suppose. In any case,
laying blame is beside the
 > point, the damage is done.
 >
 >> Especially that
 >> the topic is covered when KOffice is tagged, so
patches that "at least
 >> compile" are not enough, although thanks for
providing them - we can
 >> publish them within the release notes for users of
pqxx 2.6.9.
 >
 > How about testing them then? Tags can be slipped,
releases can be delayed.
 > There's not much point in releasing bugfix releases
with known bugs.

I consider this as packaging/release bug within pqxx. I'll
repeat this the 
last time: source compatibility is broken, not only BIC. I
am mentioning this 
not to blame anybody, this is a fact, and even surprise to
me.

I am OK with testing it, if there's someone interested. But
now any changes to 
the source can only go to Kexi 1.1.3, not 1.1.2. On Linux,
Mac, Windows, BSD. 
But unless there's someone willing to do this, Kexi supports
2.6.x series of 
pqxx, while 2.6.9 obviously is not "2.6.x series"
because the forementioned 
compatibility.

Moreover, any nontrivial changes to the source code
(assuming someone will 
want to deliver them) will break source/binary compatibility
Kexi pqxx 
driver's with "2.6.x series". I am not after
#ifdefs. It's not clear whether 
the effort will pay off, as Kexi 2.0 will most likely switch
to libpq.
Anyone is free to maintain libpqxx version then if there's a
benefit for 
additional work, it's open source... and after all Adam Pigg
has put many 
hours into this driver.

-- 
regards / pozdrawiam, Jaroslaw Staniek
  Sponsored by OpenOffice Polska (http://www.openoffice
.com.pl/en) to work on
  Kexi & KOffice: http://www.kexi-project.o
rg, http://www.koffice.org
  KDE3 & KDE4 Libraries for MS Windows: http://kdelibs.com, http://www.kde.org
_______________________________________________
Kexi-devel mailing list
Kexi-develkde.org
http
s://mail.kde.org/mailman/listinfo/kexi-devel