Re: 7.4 and Pygresql

I'm confused: Pygresql (the Python interface to Pg) was pulled from the
7.4 sources to be moved to its own project area. But, www.pygresql.org
carries a woefully out of date version of pgdb.py: it has the "ERROR:
column "typprtlen" does not exist" problem which has been fixed a few
times already, as far as I can see in the mailing list archives.
Additionally, the last updates on that site talk about the merging of
Pygresql _in_ to the Postgres sources a few years ago.

Where is Pygresql, really? Do I need to go back to the 7.3.4 (or
to-be-released-shortly 7.3.5) sources and pull it out from there? IMO,
this is quite an abominable situation for Python support; it shouldn't
have been pulled from the Postgres source tree if that really was the only
source for a current, usable version of the interface.

On December 3, 2003 10:17 pm, Arthur Ward wrote:
> I'm confused: Pygresql (the Python interface to Pg) was pulled from the
> 7.4 sources to be moved to its own project area. But, www.pygresql.org
> carries a woefully out of date version of pgdb.py: it has the "ERROR:
> column "typprtlen" does not exist" problem which has been fixed a few
> times already, as far as I can see in the mailing list archives.
> Additionally, the last updates on that site talk about the merging of
> Pygresql _in_ to the Postgres sources a few years ago.
>
> Where is Pygresql, really? Do I need to go back to the 7.3.4 (or
> to-be-released-shortly 7.3.5) sources and pull it out from there? IMO,
> this is quite an abominable situation for Python support; it shouldn't
> have been pulled from the Postgres source tree if that really was the only
> source for a current, usable version of the interface.

I will be putting out a new version very soon. In the meantime I have updated
the beta to the latest version in the repository. You can pick it up there
now. Although it is beta I have been using in pretty heavily in some
production sites and have not seen any problems.

--
D'Arcy J.M. Cain
PyGreSQL Development Group
http://www.PyGreSQL.org

Arthur Ward wrote:
> IMO,
> this is quite an abominable situation for Python support;

If you find PyGreSQL badly supported, you can use pyPgSQL or psycopg, which were
founded for this reason, among others. PyGreSQL's state was the reason I hopped on
board of the pyPgSQL development team two years ago.

> it shouldn't have been pulled from the Postgres source tree if that really was
> the only source for a current, usable version of the interface.

I suppose the GBorg project should be the definite place of the latest sources now.

Also, PyGreSQL and PoPy (yet another Python adapter for PostgreSQL which seemed
pretty badly supported to me) announced a merge of their projects a few months ago
on DB-SIG (the Python database list). My personal opinion is that there's nothing
worth keeping from the PyGreSQL codebase, and if you want a supported PostgreSQL
adapter, you should really switch to pyPgSQL or psycopg.

-- Gerhard

On December 4, 2003 03:12 am, Gerhard Häring wrote:
> Arthur Ward wrote:
> > IMO,
> > this is quite an abominable situation for Python support;
>
> If you find PyGreSQL badly supported, you can use pyPgSQL or psycopg, which
> were founded for this reason, among others. PyGreSQL's state was the reason
> I hopped on board of the pyPgSQL development team two years ago.
>
> > it shouldn't have been pulled from the Postgres source tree if that
> > really was the only source for a current, usable version of the
> > interface.
>
> I suppose the GBorg project should be the definite place of the latest
> sources now.
>
> Also, PyGreSQL and PoPy (yet another Python adapter for PostgreSQL which
> seemed pretty badly supported to me) announced a merge of their projects a
> few months ago on DB-SIG (the Python database list). My personal opinion is
> that there's nothing worth keeping from the PyGreSQL codebase, and if you
> want a supported PostgreSQL adapter, you should really switch to pyPgSQL or
> psycopg.

You know, I expect that sort of FUD from commercial enterprises that have
something to gain by switching people to their product. I am especially
disapointed to see this from you, Gerhard, after the contributions that you
have made to PygreSQL in the past.

By the way, the main reason that we moved PyGreSQL out of the main source tree
was that we could not add PyGreSQL developers to the tree without getting
them full access to the PostgreSQL tree. This move, which was fully
supported by the rest of the PostgreSQL team, was made to allow PoPy
developers in particular to have direct CVS access to the PyGreSQL tree so
that we could try to de-splinter the PostgreSQL/Python community. Given your
support in the past you would also have been welcome as a member of that
team, something I was not previously able to offer. Even so, I have always
folded your patches in. You are even mentioned on the web site as a
contributor to the 3.2 release.

PyGreSQL is far from dead. There will soon be a new web site and we will be
putting out a new version soon. Meanwhile, the CVS repository is publicly
available and the current version works quite well with the current version
of PostgreSQL. Don't write us off just yet.

--
D'Arcy J.M. Cain
PyGreSQL Development Group
http://www.PyGreSQL.org

D'Arcy J.M. Cain wrote:
> On December 4, 2003 03:12 am, Gerhard Häring wrote:
>>Also, PyGreSQL and PoPy (yet another Python adapter for PostgreSQL which
>>seemed pretty badly supported to me) announced a merge of their projects a
>>few months ago on DB-SIG (the Python database list). My personal opinion is
>>that there's nothing worth keeping from the PyGreSQL codebase, and if you
>>want a supported PostgreSQL adapter, you should really switch to pyPgSQL or
>>psycopg.
>
> You know, I expect that sort of FUD from commercial enterprises that have
> something to gain by switching people to their product.

I've recommended pyPgSQL *or* psycopg, because I honestly believe that they're
better options for developers (a lot more complete, and actively developed, and
bug reports/patches don't get lost). So it's not true that I only tried to promote
the project I'm involved in.

> I am especially disapointed to see this from you, Gerhard, after the
> contributions that you have made to PygreSQL in the past.

Yes, thanks for folding them in. I had one relevation wrt. PyGreSQL once, though,
which left a very bad taste. This was that fetchmany()/fetchone() was broken, and
was for more than one year, there apparently was a patch floating around on the
PyGreSQL list, but got lost. So when the issue came up again on pgsql-bugs, I
hunted down the bug, produced a patch and lobbied for it going in in 7.1.3 (?),
which was a task of its own.

> PyGreSQL is far from dead. There will soon be a new web site and we will be
> putting out a new version soon.

Good luck :-)

> Meanwhile, the CVS repository is publicly available and the current version
> works quite well with the current version of PostgreSQL. Don't write us off
> just yet.

Where's the new stuff?

Both the CVS at GBorg and the one at druid.net contain the PyGreSQL code I've seen
before. Nothing there looks like code from PoPy was already merged in. Is the new
codebase publicly available, yet?

It's mostly gut feeling that I don't like the current PyGreSQL implementation, but
if you want, I can try to come up with constructive criticism. This might take a
few days, though.

Let's try not to make this into a personal matter, I have nothing against you
D'Arcy and appreciate the work you've put into makeing PostgreSQL available to
Python users.

-- Gerhard

First of all I am responding using my darcy(at)druid(dot)net identity as that is the
one subscribed to these lists. Just to make it clear, Gerhard is not
responding publicly to private email. My email went to the mailing list but
I used my darcy(at)PygreSQL(dot)org identity so it is stalled waiting for moderator
approval. Perhaps, Marc, you can whitelist darcy(at)PygreSQL(dot)org as I tend to
switch to that email identity when discussing PyGreSQL.

On December 4, 2003 08:46 am, Gerhard Häring wrote:
> I've recommended pyPgSQL *or* psycopg, because I honestly believe that
> they're better options for developers (a lot more complete, and actively
> developed, and bug reports/patches don't get lost). So it's not true that I
> only tried to promote the project I'm involved in.

I appreciate that. I just think that whatever project we are involved in we
should only talk about the positives in public. This is what the PostgreSQL
project does with respect to MySQL and others. I am also a NetBSD developer
and we take the same tack wrt FreeBSD, OpenBSD, etc.

> Yes, thanks for folding them in. I had one relevation wrt. PyGreSQL once,
> though, which left a very bad taste. This was that fetchmany()/fetchone()
> was broken, and was for more than one year, there apparently was a patch
> floating around on the PyGreSQL list, but got lost. So when the issue came
> up again on pgsql-bugs, I hunted down the bug, produced a patch and lobbied
> for it going in in 7.1.3 (?), which was a task of its own.

Well, I think that we are in a better situation now. There are more people
dedicated to PyGreSQL with direct access and we have the ability to add more.
You were actually going to be on my list of people to invite to join the new
team. Would you be interested?

> > PyGreSQL is far from dead. There will soon be a new web site and we will
> > be putting out a new version soon.
>
> Good luck :-)

Thank you.

> > Meanwhile, the CVS repository is publicly available and the current
> > version works quite well with the current version of PostgreSQL. Don't
> > write us off just yet.
>
> Where's the new stuff?
>
> Both the CVS at GBorg and the one at druid.net contain the PyGreSQL code
> I've seen before. Nothing there looks like code from PoPy was already
> merged in. Is the new codebase publicly available, yet?

The GBorg system just never worked for me. I think they were having problems
at the time but I have plenty of resources to serve my own pages (including
my own ISP) so it works out just as well to use that. I should really see
about taking down the GBorg one.

As for PoPy, we are still discussing our plan of attack. We don't want to
rush in so fast that we make the tree unusable. Also, I should really get
out one more version with all the latest changes before we start the merge.
I hope to do that this weekend.

> It's mostly gut feeling that I don't like the current PyGreSQL
> implementation, but if you want, I can try to come up with constructive
> criticism. This might take a few days, though.

PyGreSQL does have that duality with the "Classic" as well as the DB-API
interface. I certainly have concentrated personally on the Classic side.
One of the things I hope to get out of this merge is a more robust DB-API
side as well as a better integration between the two. There is a lot of
redundant code in PyGreSQL. perhaps that is what your gut is complaining
about.

> Let's try not to make this into a personal matter, I have nothing against
> you D'Arcy and appreciate the work you've put into makeing PostgreSQL
> available to Python users.

I agree. I hope it is clear that I respect you and appreciate your work.
That's probably why this bothered me more than it would from someone else.

--
D'Arcy J.M. Cain
PyGreSQL Development Group
http://www.PyGreSQL.org

>>
>> You know, I expect that sort of FUD from commercial enterprises that
>> have something to gain by switching people to their product.
>
>
> I've recommended pyPgSQL *or* psycopg, because I honestly believe that
> they're better options for developers (a lot more complete, and
> actively developed, and bug reports/patches don't get lost). So it's
> not true that I only tried to promote the project I'm involved in.

It is our experience that pyPgSQL and psycopg are much better supported
than the PyGreSQL. We have tried PyGreSQL in the past and found it to be
lacking. I seriously doubt, regardless of D'Arcy's biased that he was
presenting a case based on just involvement of himself in a project.
Specifically
we have developed over 100,000 dollars worth of code in psycopg without
issue.

Sincerley,

Joshua D. Drake

--
Command Prompt, Inc., home of Mammoth PostgreSQL - S/ODBC - S/JDBC
Postgresql support, programming, shared hosting and dedicated hosting.
+1-503-222-2783 - jd(at)commandprompt(dot)com - http://www.commandprompt.com

D'Arcy J.M. Cain wrote:
> By the way, the main reason that we moved PyGreSQL out of the main
> source tree was that we could not add PyGreSQL developers to the tree
> without getting them full access to the PostgreSQL tree. This move,
> which was fully supported by the rest of the PostgreSQL team,

AFAIR, it was not even *announced* to the rest of the PostgreSQL team
until after the fact.

> was
> made to allow PoPy developers in particular to have direct CVS access

There could have been other solutions for that.

Peter Eisentraut wrote:
> D'Arcy J.M. Cain wrote:
> > By the way, the main reason that we moved PyGreSQL out of the main
> > source tree was that we could not add PyGreSQL developers to the tree
> > without getting them full access to the PostgreSQL tree. This move,
> > which was fully supported by the rest of the PostgreSQL team,
>
> AFAIR, it was not even *announced* to the rest of the PostgreSQL team
> until after the fact.

I don't remember that either. The idea of moving it out is so each
interface can have its own community and release cycle.

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

Bruce Momjian wrote:
> I don't remember that either. The idea of moving it out is so each
> interface can have its own community and release cycle.

But it doesn't take much thought to realize that 10 3-man communities
are never going to be as productive as, say, one 30-man community (Fred
Brooks notwithstanding). This community fragmentation needs to stop.

Peter Eisentraut wrote:
> Bruce Momjian wrote:
> > I don't remember that either. The idea of moving it out is so each
> > interface can have its own community and release cycle.
>
> But it doesn't take much thought to realize that 10 3-man communities
> are never going to be as productive as, say, one 30-man community (Fred
> Brooks notwithstanding). This community fragmentation needs to stop.

But people can be part of more than one community. I plan to join the
jdbc project, though I can't program in Java, but that never stopped me,
as you have seen. :-)

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

"D'Arcy J.M. Cain" <darcy(at)PyGreSQL(dot)org> writes:
> First of all I am responding using my darcy(at)druid(dot)net identity as that is the
> one subscribed to these lists. Just to make it clear, Gerhard is not
> responding publicly to private email. My email went to the mailing list but
> I used my darcy(at)PygreSQL(dot)org identity so it is stalled waiting for moderator
> approval. Perhaps, Marc, you can whitelist darcy(at)PygreSQL(dot)org as I tend to
> switch to that email identity when discussing PyGreSQL.

AFAIK there is no special whitelist for the PG mail lists. What you can
do is subscribe to the appropriate list(s) using your alternate mail
address, and then select the nomail subscription option so you don't
actually get any list traffic from that subscription. This will let
you post to the lists using either of your mail addresses.

regards, tom lane