Обсуждение: Crash at right mouse click in data grid

Hi Dave! Hi developers!

Testing pgAdmin v.1.9.0-Dev (May 2 2008, rev: 7245, on WinXP Pro; hosts:
pg 8.2.7 and 8.3.1 on Debian Etch

Looking forward to the upcoming 1.8.3 release!
You may want to have a look at the following first, though ...

I have been testing the above version occasionally over the last couple
of weeks (mostly object tree, data grid, sql window, and some random
tests). Most everything looks good or better than before. I've found a
way to crash it, though.

Using the right mouse button in the data grid (table or view) sends
pgAdmin packing almost every time. Sometimes, however, it doesn't. I
could not find why.
I have tested with a variation of tables and views in various DB
clusters, big and small, various data, various selections before right
clicking. For some time I had the hypothesis it would crash every time,
once a filter was involved. But after like 20 crashes in various
settings, it would even work with a filter on a view. Shortly after
that, right mouse click in a table (an no filter) ended the session.
So, I have no conclusive explanation to offer.

I have set pgAdmin logging to "debug". The log file invariably ends in a
couple of these lines.

2008-05-30 01:39:39 QUERY  : Scalar query (localhost:65433): SELECT
format_type(oid,NULL) as typname FROM pg_type WHERE oid = 25
2008-05-30 01:39:39 QUERY  : Query result: text
2008-05-30 01:39:39 QUERY  : Scalar query (localhost:65433): SELECT
format_type(oid,NULL) as typname FROM pg_type WHERE oid = 1114
2008-05-30 01:39:39 QUERY  : Query result: timestamp without time zone
2008-05-30 01:39:39 QUERY  : Scalar query (localhost:65433): SELECT
format_type(oid,NULL) as typname FROM pg_type WHERE oid = 23
2008-05-30 01:39:40 QUERY  : Query result: integer
2008-05-30 01:39:40 QUERY  : Scalar query (localhost:65433): SELECT
format_type(oid,NULL) as typname FROM pg_type WHERE oid = 16
2008-05-30 01:39:40 QUERY  : Query result: boolean

I can send some sample log files if that helps.


Regards
Erwin

Hi Erwin

On Fri, May 30, 2008 at 1:14 AM, Erwin Brandstetter
<brandstetter@falter.at> wrote:
> Hi Dave! Hi developers!
>
> Testing pgAdmin v.1.9.0-Dev (May 2 2008, rev: 7245, on WinXP Pro; hosts: pg
> 8.2.7 and 8.3.1 on Debian Etch
>
> Looking forward to the upcoming 1.8.3 release!
> You may want to have a look at the following first, though ...

It's a bug in a new feature, so won't affect 1.8.3 (thankfully :-p )

> I have been testing the above version occasionally over the last couple of
> weeks (mostly object tree, data grid, sql window, and some random tests).
> Most everything looks good or better than before. I've found a way to crash
> it, though.
>
> Using the right mouse button in the data grid (table or view) sends pgAdmin
> packing almost every time. Sometimes, however, it doesn't. I could not find
> why.

It's crashing in a thread->IsRunning() test in some new code that
Robins added to display a popup menu of sorting and filtering options.
It's not clear to me why it's crashing though - the code looks as if
it should work just fine. Can you take a look in case I missed
something please Robins?

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

On Sat, May 31, 2008 at 12:46 AM, Robins Tharakan <tharakan@gmail.com> wrote:

> In particular, the bug report in the April conversation, said that the crash
> is due to IsRunning() check that Erwin is pointing to.

Oh yeah, that one :-(. I'm beginning to think we could use a bug
tracker (and I hate Bugzilla before anyone suggest it!).

I'm still stumped about that now, but I've worked around it by
checking if the refresh button is enabled as we already do in a few
other similar places. The data refresh code enables/disables that for
safety so it should be a reliable indicator of whether or not a thread
is running.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page a écrit :
> On Sat, May 31, 2008 at 12:46 AM, Robins Tharakan <tharakan@gmail.com> wrote:
>
>> In particular, the bug report in the April conversation, said that the crash
>> is due to IsRunning() check that Erwin is pointing to.
>
> Oh yeah, that one :-(. I'm beginning to think we could use a bug
> tracker (and I hate Bugzilla before anyone suggest it!).
>

trac would be a good one : wiki, bug tracker and source browser. Seems a
good deal :)

> I'm still stumped about that now, but I've worked around it by
> checking if the refresh button is enabled as we already do in a few
> other similar places. The data refresh code enables/disables that for
> safety so it should be a reliable indicator of whether or not a thread
> is running.
>


--
Guillaume.
  http://www.postgresqlfr.org
  http://dalibo.com

On Sun, Jun 8, 2008 at 9:12 PM, Guillaume Lelarge
<guillaume@lelarge.info> wrote:

>
> trac would be a good one : wiki, bug tracker and source browser. Seems a
> good deal :)

We use trac for the postgresql.org infrastructure and website
management. It seems to go wrong on a fairly regular basis :-(

Oh, and I just found this literally 30 minutes ago whilst adding some
data to the PostgreSQL product catalogue I've been hacking on:
http://www.initd.org/ (that's the psycopg site).

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com

Dave Page wrote:
> On Sun, Jun 8, 2008 at 9:12 PM, Guillaume Lelarge
> <guillaume@lelarge.info> wrote:
>
>> trac would be a good one : wiki, bug tracker and source browser. Seems a
>> good deal :)
>
> We use trac for the postgresql.org infrastructure and website
> management. It seems to go wrong on a fairly regular basis :-(

That really isn't Trac's fault...


> Oh, and I just found this literally 30 minutes ago whilst adding some
> data to the PostgreSQL product catalogue I've been hacking on:
> http://www.initd.org/ (that's the psycopg site).

I'd be happy to relay some internal info on that specific case you to on
IM, if you're interested, but it's not something I'll post in public.


//Magnus

Dave Page a écrit :
> On Sun, Jun 8, 2008 at 9:12 PM, Guillaume Lelarge
> <guillaume@lelarge.info> wrote:
>
>> trac would be a good one : wiki, bug tracker and source browser. Seems a
>> good deal :)
>
> We use trac for the postgresql.org infrastructure and website
> management. It seems to go wrong on a fairly regular basis :-(
>

I know the webteam used it but wasn't aware of any problems.

> Oh, and I just found this literally 30 minutes ago whilst adding some
> data to the PostgreSQL product catalogue I've been hacking on:
> http://www.initd.org/ (that's the psycopg site).
>

Hehe. First time I heard of problems with trac. So, what about mantis ?
dalibo uses it and it works great (yeah, I know, I can also tell the
same about trac... PostgreSQLfr use it for the translation of the
PostgreSQL manual).


--
Guillaume.
  http://www.postgresqlfr.org
  http://dalibo.com

On Sun, Jun 8, 2008 at 9:26 PM, Guillaume Lelarge
<guillaume@lelarge.info> wrote:
> Dave Page a écrit :
>>
>> On Sun, Jun 8, 2008 at 9:12 PM, Guillaume Lelarge
>> <guillaume@lelarge.info> wrote:
>>
>>> trac would be a good one : wiki, bug tracker and source browser. Seems a
>>> good deal :)
>>
>> We use trac for the postgresql.org infrastructure and website
>> management. It seems to go wrong on a fairly regular basis :-(
>>
>
> I know the webteam used it but wasn't aware of any problems.

Yeah - apparently it's got something to do with having 2 instances on
the same box. If it's not Trac's fault (as Magnus asserts), then it
must be the fault of one of it's dependencies, which is just a bad
imho.

>> Oh, and I just found this literally 30 minutes ago whilst adding some
>> data to the PostgreSQL product catalogue I've been hacking on:
>> http://www.initd.org/ (that's the psycopg site).
>>
>
> Hehe. First time I heard of problems with trac. So, what about mantis ?
> dalibo uses it and it works great (yeah, I know, I can also tell the same
> about trac... PostgreSQLfr use it for the translation of the PostgreSQL
> manual).

No idea about Mantis. Whatever we use, it should be as simple as
possible imho. I'd need to look into it some more.

--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com