Thread: Snapshot release 08.00.0002 uploaded

Snapshot release 08.00.0002 uploaded

From
"Dave Page"
Date:
I've just uploaded a new snapshot to
ftp://ftp.postgresql.org/pub/odbc/versions/snapshots/ (it should hit the
mirrors within 24 hours). This version of the driver is a minor update
with a fix to the string sizing code in SQLGetInfo, and removal of the
need for tables to be in the search path or public schema for them to be
returned by SQLTables.

Feedback would be appreciated :-)

Regards, Dave.

Re: Snapshot release 08.00.0002 uploaded

From
Giulio Orsero
Date:
On Fri, 8 Oct 2004 21:45:26 +0100, "Dave Page" <dpage@vale-housing.co.uk>
wrote:

>I've just uploaded a new snapshot to
>ftp://ftp.postgresql.org/pub/odbc/versions/snapshots/ (it should hit the
>mirrors within 24 hours). This version of the driver is a minor update
>with a fix to the string sizing code in SQLGetInfo, and removal of the
>need for tables to be in the search path or public schema for them to be
>returned by SQLTables.

Since you are in a development phase, can you tell me if the following
feature is in your roadmap?

Add a check box "Quote every table/field identifier" yes/no, default: no

This would solve one of the greatest problems interacting with psql server:
- On the server you define MyTable.MyField
- Excel/MSQuery receives MyTable.MyField names through ODBC
- Excel/MSQuery builds a query with MyTable.MyField (not
"MyTable"."MyField")
- psql server lowercase to mytable.myfield and errors out

I read all the explanations about how this behavior is right/good, but users
just want it to work.

Thanks

--
giulioo@pobox.com

Re: Snapshot release 08.00.0002 uploaded

From
"Dave Page"
Date:

> -----Original Message-----
> From: pgsql-odbc-owner@postgresql.org
> [mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of Giulio Orsero
> Sent: 08 October 2004 22:11
> To: pgsql-odbc@postgresql.org
> Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>
> Since you are in a development phase, can you tell me if the
> following feature is in your roadmap?

I'm in more of a 'get it stable for release' mode to be honest.

> Add a check box "Quote every table/field identifier" yes/no,
> default: no

No it's not. This is actually pretty difficult to do because it requires
parsing and rewriting queries that are generated externally. If you log
a feature request on the Gborg site however, that'll ensure it doesn't
get forgotten.

Regards, Dave.

Re: Snapshot release 08.00.0002 uploaded

From
Karol Szkudlarek
Date:
Dave Page wrote:
> I've just uploaded a new snapshot to
> ftp://ftp.postgresql.org/pub/odbc/versions/snapshots/ (it should hit the
> mirrors within 24 hours). This version of the driver is a minor update
> with a fix to the string sizing code in SQLGetInfo, and removal of the
> need for tables to be in the search path or public schema for them to be
> returned by SQLTables.
>
> Feedback would be appreciated :-)
>
> Regards, Dave.
>
> ---------------------------(end of broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
>
>                http://www.postgresql.org/docs/faqs/FAQ.html
>
Hi!

I repost my previous e-mail. I checked the shapshot and the follwing
problem still exists:


Karol Szkudlarek wrote:

 >
 > Hi!
 >
 > My friend have problem with latest snapshot of psql-odbc driver.
 > He doesn't subsribe to the mailing list, so I forwarded his mail.
 > Please help him... :-)
 >
 > -------- Original Message --------
 > Subject: [Fwd: SQLDescribeCol before SQLExcecute call problerm]
 > Date: Mon, 20 Sep 2004 09:50:29 +0200 (CEST)
 > From: mik@mikronika.com.pl
 > To: karol@mikronika.com.pl
 >
 > -------------------------- Wiadomość oryginalna
--------------------------
 > Temat: SQLDescribeCol before SQLExcecute call problerm
 > Od:    mik@mikronika.com.pl
 > Data:  Pt Września 17 2004, 11:32
 > Do:    pgsql-odbc@postgresql.org
 >
--------------------------------------------------------------------------
 >
 >
 > Hi!
 >
 > I have problem with the latest odbc postgres driver. In previous version
 > of the driver I can call SQLDescribeCol and SQLExecute in any
 > order. Now if I call SQLDescribeCol before SQLExecute I got error:
 >
 > HY010 (Function sequence error). ODBC SDK documentation allows calling in
 > that order. Is it the feature or bug?
 >
 > I have many lines of code depend on library (OTL) which using that
sort of
 > calling convention. And now that will be very difficult to change it.
 >
 > Problem concerns both the latest snapshot and cvs-tree on Windows and
Linux.
 >
 > Regards,
 >
 > Michał Konopiński
 >

Re: Snapshot release 08.00.0002 uploaded

From
"Dave Page"
Date:

> -----Original Message-----
> From: Karol Szkudlarek [mailto:karol@mikronika.com.pl]
> Sent: 11 October 2004 15:21
> To: Dave Page
> Cc: pgsql-odbc@postgresql.org
> Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>
>
>
> I repost my previous e-mail. I checked the shapshot and the
> follwing problem still exists:

Err, yes. I only worked on SQLInfo and SQLTables as I said.

What previous version of the driver did this sequence of calls work in?

Regards, Dave


Re: Snapshot release 08.00.0002 uploaded

From
Karol Szkudlarek
Date:
Dave Page wrote:
>
>
>
>>-----Original Message-----
>>From: Karol Szkudlarek [mailto:karol@mikronika.com.pl]
>>Sent: 11 October 2004 15:21
>>To: Dave Page
>>Cc: pgsql-odbc@postgresql.org
>>Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>>
>>
>>
>>I repost my previous e-mail. I checked the shapshot and the
>>follwing problem still exists:
>
>
> Err, yes. I only worked on SQLInfo and SQLTables as I said.
>
> What previous version of the driver did this sequence of calls work in?
>
> Regards, Dave
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
>

Works only with the stable version of the driver. Any snapshot version
doesn't work.

Grests, Karol

Re: Snapshot release 08.00.0002 uploaded

From
"Dave Page"
Date:

> -----Original Message-----
> From: pgsql-odbc-owner@postgresql.org
> [mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of Karol Szkudlarek
> Sent: 11 October 2004 21:16
> To: pgsql-odbc@postgresql.org
> Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>
> Works only with the stable version of the driver. Any
> snapshot version doesn't work.

Hi,

I took a look at this, but to be honest am not sure what might be
causing it. Unfortunately I don't have time right now to spend trying to
understand how those bits of the code work.

Can you log this on the Gborg project bugs list so it doesn't get
forgotten please?

Thanks, Dave

Re: Snapshot release 08.00.0002 uploaded

From
Karol Szkudlarek
Date:
>>-----Original Message-----
>>From: pgsql-odbc-owner@postgresql.org
>>[mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of Karol Szkudlarek
>>Sent: 11 October 2004 21:16
>>To: pgsql-odbc@postgresql.org
>>Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>>
>>Works only with the stable version of the driver. Any
>>snapshot version doesn't work.
>
>
> Hi,
>
> I took a look at this, but to be honest am not sure what might be
> causing it. Unfortunately I don't have time right now to spend trying to
> understand how those bits of the code work.
>
> Can you log this on the Gborg project bugs list so it doesn't get
> forgotten please?
>
> Thanks, Dave
>

Hi!

Sorry for delay... Today at the morning I submitted bug on gbord. The
bug ID is 1012.

Greets,
Karol

Re: Snapshot release 08.00.0002 uploaded

From
Max Cohan
Date:
You might want to submit this to the main bug/wish list.
This seems like something sane and probably fairly easy to support
in the server itself. It seems like there is nothing implicitly wrong
with the "table"."field" syntax and the server itself should understand
it.

Max

On Fri, Oct 08, 2004 at 11:11:29PM +0200, Giulio Orsero wrote:
> On Fri, 8 Oct 2004 21:45:26 +0100, "Dave Page" <dpage@vale-housing.co.uk>
> wrote:
>
> >I've just uploaded a new snapshot to
> >ftp://ftp.postgresql.org/pub/odbc/versions/snapshots/ (it should hit the
> >mirrors within 24 hours). This version of the driver is a minor update
> >with a fix to the string sizing code in SQLGetInfo, and removal of the
> >need for tables to be in the search path or public schema for them to be
> >returned by SQLTables.
>
> Since you are in a development phase, can you tell me if the following
> feature is in your roadmap?
>
> Add a check box "Quote every table/field identifier" yes/no, default: no
>
> This would solve one of the greatest problems interacting with psql server:
> - On the server you define MyTable.MyField
> - Excel/MSQuery receives MyTable.MyField names through ODBC
> - Excel/MSQuery builds a query with MyTable.MyField (not
> "MyTable"."MyField")
> - psql server lowercase to mytable.myfield and errors out
>
> I read all the explanations about how this behavior is right/good, but users
> just want it to work.
>
> Thanks
>
> --
> giulioo@pobox.com
>
> ---------------------------(end of broadcast)---------------------------
> TIP 8: explain analyze is your friend

Re: Snapshot release 08.00.0002 uploaded

From
"Dave Page"
Date:

> -----Original Message-----
> From: pgsql-odbc-owner@postgresql.org
> [mailto:pgsql-odbc-owner@postgresql.org] On Behalf Of Max Cohan
> Sent: 21 October 2004 07:49
> To: pgsql-odbc@postgresql.org
> Subject: Re: [ODBC] Snapshot release 08.00.0002 uploaded
>
>
> You might want to submit this to the main bug/wish list.
> This seems like something sane and probably fairly easy to
> support in the server itself. It seems like there is nothing
> implicitly wrong with the "table"."field" syntax and the
> server itself should understand it.

It does. What it doesn't understand is mixed case unquoted identifiers,
which other databases should also barf on - the difference being,
PostgreSQL folds them to lower case, other DBMSs fold them to upper
case.

Regards, Dave

08.00.0002 useless :-)

From
Johann Zuschlag
Date:
Hello,

somebody already mentioned it. I just want to repeat it. As long as the
driver produces strings like this:

CREATE TABLE FOO (Userid varchar()(10) NOT NULL, Mcode varchar()(25)...

it is unfortunately useless. The backend will never accept it.

I'm sure Hiroshi had something different in mind when he changed that
part of the code.

Does somebody work on that?

Regards

Johann


ODBC Driver stability

From
Gary Doades
Date:
Unfortunately I have to come to the conclusion that the latest ODBC
driver (8.0.0.2) is not as stable as first thought. When I initially
tested it I was using a single application with a few connections and it
seemed OK (better than 8.0.0.1 anyway)

Recently I have been doing heavier load testing within my ASP.NET
application. Being a Web app on Windows it is inheriently multi-threaded
and uses the MS .NET connect-execute-fetch-disconnect cycle heavily.

After obtaining just 10 connections and executing SQL the app starts to
report random database failure messages "No error" and null reference
exceptions withing Syetem.Data are common.

Various restarts, checking there are no other psqlodbc dlls and various
test loads end up with the same situation fairly quickly.

Switching back to the 7.3.2.0 ODBC driver and the app will run stable
for hours (or for as long as the test runs) with no driver errors.

I can now reproduce this problem in a single Windows GUI app by getting
lots of connections and radpidly executing SQL immediately after opening
each of the connections.
Sadly being in the .NET environment I can't see how to debug the driver
in these situations or how to get more information about the faults. It
still looks like buffer overruns causing the main problems.

I'm a bit concerned that Postgres is going to end up with a fantastic
server (even on Windows!), but badly let down by the ODBC driver, which
I suspect that a lot of the Windows newcomers will want to use. I know
that 7.3.2.0 is stable at the moment, but what is the way forward?

Regards,
Gary.

Re: ODBC Driver stability

From
"Markus Donath"
Date:
Hello,

i have also tested several PGSQL ODBC driver versions with Windows 2000, IIS
(the Windows webserver) and ASP.
Version 7.3.2.0 runs best in a multithreaded environment, but it is still
not 100% stable.
The webserver sometimes hangs (depends on the load) so that the process
needs to be killed.
All in all, the Postgres ODBC driver does not fit our needs for stability,
and as the posts in this newsgroup show, there is no development going on.
To me, it looks like no one here uses the ODBC-driver for server
applications with windows.
Our conclusion is that Postgres is not usable with ODBC on windows, and no
stable driver will be available soon. We are going to switch a different
database. To bad, as the database itself runs fast and stable.

bye bye postgres !


http://home.arcor.de/mkdna/


> -----Ursprüngliche Nachricht-----
> Von: pgsql-odbc-owner@postgresql.org
> [mailto:pgsql-odbc-owner@postgresql.org]Im Auftrag von Gary Doades
> Gesendet: Montag, 8. November 2004 23:30
> An: pgsql-odbc@postgresql.org
> Betreff: [ODBC] ODBC Driver stability
>
>
> Unfortunately I have to come to the conclusion that the latest ODBC
> driver (8.0.0.2) is not as stable as first thought. When I initially
> tested it I was using a single application with a few connections and it
> seemed OK (better than 8.0.0.1 anyway)
>
> Recently I have been doing heavier load testing within my ASP.NET
> application. Being a Web app on Windows it is inheriently multi-threaded
> and uses the MS .NET connect-execute-fetch-disconnect cycle heavily.
>
> After obtaining just 10 connections and executing SQL the app starts to
> report random database failure messages "No error" and null reference
> exceptions withing Syetem.Data are common.
>
> Various restarts, checking there are no other psqlodbc dlls and various
> test loads end up with the same situation fairly quickly.
>
> Switching back to the 7.3.2.0 ODBC driver and the app will run stable
> for hours (or for as long as the test runs) with no driver errors.
>
> I can now reproduce this problem in a single Windows GUI app by getting
> lots of connections and radpidly executing SQL immediately after opening
> each of the connections.
> Sadly being in the .NET environment I can't see how to debug the driver
> in these situations or how to get more information about the faults. It
> still looks like buffer overruns causing the main problems.
>
> I'm a bit concerned that Postgres is going to end up with a fantastic
> server (even on Windows!), but badly let down by the ODBC driver, which
> I suspect that a lot of the Windows newcomers will want to use. I know
> that 7.3.2.0 is stable at the moment, but what is the way forward?
>
> Regards,
> Gary.
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
>
>
> ---
> Incoming mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.789 / Virus Database: 534 - Release Date: 07.11.2004
>


Re: ODBC Driver stability

From
Shachar Shemesh
Date:
Markus Donath wrote:

>Hello,
>
>Our conclusion is that Postgres is not usable with ODBC on windows, and no
>stable driver will be available soon.
>
Have you tried other options? There are both a .NET driver and an OleDB
provider, though can only attest to the second.

OLE DB is nearing the point where it should be quite usable for server
use. The .NET provider has been around even longer, so may be even
better. In any case, the OLE infrastructure prevents threading problems
(the hard way, unfortunately), and PgOleDb uses libpq for actual
database communication.

> We are going to switch a different
>database. To bad, as the database itself runs fast and stable.
>
>bye bye postgres !
>
>
Sad to see you go.

          Shachar

--
Shachar Shemesh
Lingnu Open Source Consulting ltd.
http://www.lingnu.com/


Re: ODBC Driver stability

From
Date:
Hello Shachar,

oh ... good news ... i didn't know that the ole db driver has been released.
i will check the ole db driver, thank you for the information !

what do you mean with "OLE ... prevents threading problems the hard way" ?

bye,
Markus.

> -----Ursprüngliche Nachricht-----
> Von: Shachar Shemesh [mailto:psql@shemesh.biz]
> Gesendet: Dienstag, 16. November 2004 14:43
> An: Markus Donath
> Cc: Gary Doades; pgsql-odbc@postgresql.org
> Betreff: Re: [ODBC] ODBC Driver stability
>
>
> Markus Donath wrote:
>
> >Hello,
> >
> >Our conclusion is that Postgres is not usable with ODBC on
> windows, and no
> >stable driver will be available soon.
> >
> Have you tried other options? There are both a .NET driver and an OleDB
> provider, though can only attest to the second.
>
> OLE DB is nearing the point where it should be quite usable for server
> use. The .NET provider has been around even longer, so may be even
> better. In any case, the OLE infrastructure prevents threading problems
> (the hard way, unfortunately), and PgOleDb uses libpq for actual
> database communication.
>
> > We are going to switch a different
> >database. To bad, as the database itself runs fast and stable.
> >
> >bye bye postgres !
> >
> >
> Sad to see you go.
>
>           Shachar
>
> --
> Shachar Shemesh
> Lingnu Open Source Consulting ltd.
> http://www.lingnu.com/
>
>
>
> ---
> Incoming mail is certified Virus Free.
> Checked by AVG anti-virus system (http://www.grisoft.com).
> Version: 6.0.796 / Virus Database: 540 - Release Date: 13.11.2004
>


Re: ODBC Driver stability

From
Shachar Shemesh
Date:
md-postgres-donath@netapps.de wrote:

>Hello Shachar,
>
>oh ... good news ... i didn't know that the ole db driver has been released.
>i will check the ole db driver, thank you for the information !
>
>
It has not been "officially" released yet, but version 1.0.0.15 made a
major move in that direction. It has purged itself from ugly debug
asserts and message boxes. In addition to that, it is being actively
maintained, and so I'm hoping for a proper release soon.

>what do you mean with "OLE ... prevents threading problems the hard way" ?
>
>
OLE (well, DCOM really) allows a threading solution for the lazy (for
the time being, that's us). The server runs (by default) in a single
thread, and all other threads use a proxy to call it. Right now we are
too focused on getting this version functionally complete to start
worrying about this. As OLE DB is generally a faster concept anyways,
I'm hoping this won't be too much of a problem.

Work on getting multi threading into the provider cannot start until I
understand libpq's thread safety.

          Shachar

--
Shachar Shemesh
Lingnu Open Source Consulting ltd.
http://www.lingnu.com/