Re: What I would say if someone asked me about no

Bruce Momjian wrote:

> I ran out of time, just like the PITR guy. Win32 takes tons of time,
> and working on the PostgreSQL project itself just doesn't allow enough
> time.

It is also partly my fault, as I failed on the first attempt to redo the
fork+exec part for Bruce, then got a bit frustrated over things not to
be discussed publicly and didn't give it much attention, finally I had a
ton of other things with higher priority after that. This is not an
excuse, it's life, sorry.

What really bugs me about the entire Win32 port discussion is something
else. One of my activities at PeerDirect was to extract out the 7.2.1
based Win32 port and contribute it as a patch. That patch went out on
January 20, 2003. Applied to a 7.2.1 tree, it compiled under MS VC++ 6.0
and produced a fully running native Win32 PostgreSQL. So far for what
was available in January.

Now more than half a year later we have no evidence that there will be
qualified and capable maintainers for any Win32 port. Some of the main
developers who implement big features and do the kind of work that can
break a port easily, especially one that serves a tremendeously
different API, have explicitly waved off any interest in Win32 support
at all. This is a dangerous combination.

Bruce and I had discussed this issue briefly and he is of course right
that this sort of task requires deep PostgreSQL and Win32 knowledge,
both of which takes time to build up. It's the kind of task that
requires Core developer help to have any chance. Where I disagree by now
is the definition of "help". Just doing the hard parts is not the kind
of help needed here. If we do not get people growing into this position
right now, there will be a gap between when the port is once done and
when others with real Win32 interest will take over the duties. Will
there ever be anyone willing to take over an already broken port? And I
am sure if nobody is really committed on maintaining this port, it'll be
broken in a few months.

At this point I (personally) would rather not have a Win32 port at all,
because all it is buying us under these circumstances will be bad
reputation. We have come a long way and the fact that people do not
reevaluate things is the reason they still "know" that PostgreSQL is
slow and instable. Adding a bad maintained Win32 port is worse than not
supporting Win32 at all. The latter might be missing an important
platform, the former will build up long lasting bad reputation.

Jan

>
> ---------------------------------------------------------------------------
>
> Josh Berkus wrote:
>> People,
>>
>> > Q. Hey, I read in a bunch of places that 7.4 was going to have native
>> > windows support, what gives?
>>
>> My answer:
>>
>> A. PostgreSQL has a well-deserved reputation for "bulletproof" reliability and
>> stability, which requires each new patch and plug-in to pass stringent
>> community testing. The Win32 platform has proved to be more of a challenge
>> than we anticipated in this regard, and we would rather release our Windows
>> port late than release a version which might fail in production environments.
>>
>> But, Bruce, since you're head of the port, wanna comment on this? We're
>> trying to prepare a "canned" answer for the inevitable questions.
>>
>> --
>> Josh Berkus
>> Aglio Database Solutions
>> San Francisco
>>
>> ---------------------------(end of broadcast)---------------------------
>> TIP 1: subscribe and unsubscribe commands go to majordomo(at)postgresql(dot)org
>>
>

--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck(at)Yahoo(dot)com #