Thread: FW: oids bug

FW: oids bug

From
"Dave Page"
Date:
Michael,

I have forwarded this to the support mailing list. Please use the lists
rather than private email as I am not necessarily the person to help
with these problems.

Regards, Dave.

> -----Original Message-----
> From: Michael Shapiro [mailto:mshapiro@ncsa.edu]
> Sent: 25 November 2003 19:55
> To: Dave Page
> Subject: oids bug
>
> Dave, I re-created all mytables with OIDs. Now the View Data
> allows me to edit the tables. However, PgAmin crashes when I
> make changes. I think you mentioned that this might be the case......
>
> ---
> Michael
>
>


Re: FW: oids bug

From
Andreas Pflug
Date:
>
>  
>
>>-----Original Message-----
>>From: Michael Shapiro [mailto:mshapiro@ncsa.edu] 
>>Sent: 25 November 2003 19:55
>>To: Dave Page
>>Subject: oids bug
>>
>>Dave, I re-created all mytables with OIDs. Now the View Data 
>>allows me to edit the tables. However, PgAmin crashes when I 
>>make changes. I think you mentioned that this might be the case......
>>
>>    
>>
Michael,

can you give some details? I can't reproduce this on my tables.

Regards,
Andreas



Re: FW: oids bug

From
Andreas Pflug
Date:
Michael Shapiro wrote:

> I forgot to send you the trigger for this table. It turns out that the
> trigger
> raises an exception when an update on the packet_id occurs.
>
> At 07:34 PM 11/26/2003 +0100, you wrote:
>
>> Michael Shapiro wrote:
>>
>>> NOTE, I am running Windows XP, PgAdmin 1.1.0 Devel(Nov 21 2003)
>>
>>
>> win32 version shouldn't make any difference. There has been a similar
>> observation some days ago, which disappeared after recreation of the
>> table (which sounds like magic to me). I'd be glad if we could find
>> the problem; pgAdmin3 may not crash whatever data should be in the
>> backend.
>

Finally, I found it.
After executing a query in a thread, the pointer to the notice
processing routine was dangling. An update wouldn't execute in a thread,
but directly, receiving a NOTICE into the wild.

New version committed to cvs, win32 binary upload to follow.
This might fix the Edit Data problem previously reported by Jonathan
Mezach too.

Regards,
Andreas