Обсуждение: SQL Editor fail

Hi! This is my first message in list, I hope that it will be helpful.

I'm finding a bug in pgAdmin 1.14.0 Beta 1.
To make sure that the problem is still there I'm pulled from latest git today.

To reproduce it, enter in SQL Editor, write symbol "(" and then press
enter for go to new line. It crashing pgAdmin.

I'm testing it on MacOS 10.6.7.

Regards,
Aleksey

On Thu, May 12, 2011 at 1:15 PM, Aleksey Chirkin <a4irkin@gmail.com> wrote:
> Hi! This is my first message in list, I hope that it will be helpful.
>
> I'm finding a bug in pgAdmin 1.14.0 Beta 1.
> To make sure that the problem is still there I'm pulled from latest git today.
>
> To reproduce it, enter in SQL Editor, write symbol "(" and then press
> enter for go to new line. It crashing pgAdmin.
>
> I'm testing it on MacOS 10.6.7.

Hmm, I can reproduce it, but I have no idea what the problem is. The
stack trace (which contains 56977 frames!!) indicates the crash
happens deep in Apples rendering code, which in turn is called from
wxWidgets painting code which is called by the main event loop. As far
as I can see, there is no pgAdmin code in the stack at all.

It seems to occur any time you hit return when there's an unmatched
bracket of any type on the line. I don't think the block folding is
the issue, as the crash doesn't occur with text like "BEGIN" which
will also start a new code block.

-- 
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

On 12 May 2011 16:15, Dave Page <dpage@pgadmin.org> wrote:

> Hmm, I can reproduce it, but I have no idea what the problem is. The
> stack trace (which contains 56977 frames!!) indicates the crash
> happens deep in Apples rendering code, which in turn is called from
> wxWidgets painting code which is called by the main event loop. As far
> as I can see, there is no pgAdmin code in the stack at all.

Yeah, I tend to see that a lot with GUI code, where events are handled
asynchronously. Often, the trick is to carpetbomb a whole bunch of
breakpoints haphazardly in the general area of the problem, and then
when triggering the problem, as each breakpoint is hit, disable them
one-by-one, until you see the error again, at which point you
re-enable the most recently disabled breakpoint. You are now near to
the problem.

Sorry if that's obvious - I can't really tell.

--
Peter Geoghegan       http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training and Services

Le 05/12/2011 02:15 PM, Aleksey Chirkin a écrit :
> Hi! This is my first message in list, I hope that it will be helpful.
> 
> I'm finding a bug in pgAdmin 1.14.0 Beta 1.
> To make sure that the problem is still there I'm pulled from latest git today.
> 
> To reproduce it, enter in SQL Editor, write symbol "(" and then press
> enter for go to new line. It crashing pgAdmin.
> 
> I'm testing it on MacOS 10.6.7.
> 

Works on Linux at least. As Dave is able to reproduce it, I created a
ticket (http://code.pgadmin.org/trac/ticket/316) on this issue so that
we can work on it.

Thanks for your report.


-- 
Guillaumehttp://www.postgresql.frhttp://dalibo.com

On Thu, May 12, 2011 at 3:15 PM, Dave Page <dpage@pgadmin.org> wrote:
> On Thu, May 12, 2011 at 1:15 PM, Aleksey Chirkin <a4irkin@gmail.com> wrote:
>> Hi! This is my first message in list, I hope that it will be helpful.
>>
>> I'm finding a bug in pgAdmin 1.14.0 Beta 1.
>> To make sure that the problem is still there I'm pulled from latest git today.
>>
>> To reproduce it, enter in SQL Editor, write symbol "(" and then press
>> enter for go to new line. It crashing pgAdmin.
>>
>> I'm testing it on MacOS 10.6.7.
>
> Hmm, I can reproduce it, but I have no idea what the problem is. The
> stack trace (which contains 56977 frames!!) indicates the crash
> happens deep in Apples rendering code, which in turn is called from
> wxWidgets painting code which is called by the main event loop. As far
> as I can see, there is no pgAdmin code in the stack at all.
>
> It seems to occur any time you hit return when there's an unmatched
> bracket of any type on the line. I don't think the block folding is
> the issue, as the crash doesn't occur with text like "BEGIN" which
> will also start a new code block.

I was finally able to figure this out, and have committed a fix for it.

Thanks for the report!

-- 
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company