Re: Latest SVN segfaults on Feisty

I finally managed to build from SVN on Ubuntu Feisty after the
"expected specifier-qualifier-list before 'JoinType'" problem was
fixed. Unfortunately it segfaults now. Anyone care to see strace /
ltrace? I've put them here: http://public.box.net/ivanpavlov

ivan(dot)pavlov(at)gmail(dot)com wrote:
> I finally managed to build from SVN on Ubuntu Feisty after the
> "expected specifier-qualifier-list before 'JoinType'" problem was
> fixed. Unfortunately it segfaults now. Anyone care to see strace /
> ltrace? I've put them here: http://public.box.net/ivanpavlov

Would you mind mailing the trace in a more standard format please? I'd
rather not have to go find something to open rar files.

Regards, Dave

Ivan Pavlov wrote:
> Oops, you must be using Windows. Please find the attached traces.

Mac and Windows actually. Thanks - I'll take a look.

Regards, Dave

Ivan Pavlov wrote:
> Oops, you must be using Windows. Please find the attached traces.

Hmm, thats a trace from hell! What are you doing exactly when it crashes?

Regards, Dave.

On May 18, 4:03 pm, d(dot)(dot)(dot)(at)postgresql(dot)org (Dave Page) wrote:
> Hmm, thats a trace from hell! What are you doing exactly when it crashes?

I don't do anything out of the ordinary - I have a couple of Java apps
up and running, perhaps some extra activity on a lighttpd server, two
postgresql instances - 8.1 and 8.2.

What exactly do you need to know?

Ivan

ivan(dot)pavlov(at)gmail(dot)com wrote:
> On May 18, 4:03 pm, d(dot)(dot)(dot)(at)postgresql(dot)org (Dave Page) wrote:
>> Hmm, thats a trace from hell! What are you doing exactly when it crashes?
>
> I don't do anything out of the ordinary - I have a couple of Java apps
> up and running, perhaps some extra activity on a lighttpd server, two
> postgresql instances - 8.1 and 8.2.
>
> What exactly do you need to know?

When does it crash? At startup? When you try to add a new constraint?
When you connect to a server? etc. etc.

Regards, Dave.

It crashes at startup. Btw, I have 1.4.3 from a package and it works fine.
Another piece of info - the build from svn was working fine until perhaps
the beginning of April, then I couldn't compile anymore for some time. Now
it crashes. Hope that helps.

Ivan

When does it crash? At startup? When you try to add a new constraint?
> When you connect to a server? etc. etc.
>
> Regards, Dave.
>
>

Ivan Pavlov wrote:
> It crashes at startup. Btw, I have 1.4.3 from a package and it works
> fine. Another piece of info - the build from svn was working fine until
> perhaps the beginning of April, then I couldn't compile anymore for some
> time. Now it crashes. Hope that helps.

OK, you're using SVN code - I assume you've updated? Please do so again
(I've committed changes this morning), then run the bootstrap script,
followed by 'make clean; make all'.

Then, edit ~/.pgadmin3, and set (or add) the following values:

LogFile=/some/path/you/can/write/to/pgadmin.log
LogLevel=4

Finally, start pgAdmin. This should create a logfile which will
hopefully give some clues about what's going on.

Regards, Dave.

Ivan Pavlov wrote:
> I re-compiled and set LogLevel to 4 but there is nothing new written in
> the log file.

I haven't seen a pgadmin logfile from you yet so I wouldn't know what is
new and what isn't!

> I attach the trace files again, just in case there is some
> difference.

They don't help much unfortuantely. Can you fire it up in gdb and get a
plain old back trace please?

Thanks, Dave

On 5/18/07, Dave Page <dpage(at)postgresql(dot)org> wrote:
>
> I haven't seen a pgadmin logfile from you yet so I wouldn't know what is
> new and what isn't!

What I meant is - all entries in the log are timestamped and there were
only some month old sql errors in there.

They don't help much unfortuantely. Can you fire it up in gdb and get a
> plain old back trace please?

Here it is:
#0 0xb77225ae in wxStringBase::find () from /usr/lib/libwx_baseu-2.8.so.0
#1 0xb772534a in wxStringBase::find () from /usr/lib/libwx_baseu-2.8.so.0
#2 0xb77253e4 in wxString::Find () from /usr/lib/libwx_baseu-2.8.so.0
#3 0x08347dee in isPgApp ()
#4 0x0807ace5 in pgAdmin3::InitPaths ()
#5 0x0808168e in pgAdmin3::OnInit ()
#6 0x08084d01 in wxAppConsole::CallOnInit ()
#7 0xb7700250 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#8 0xb7700327 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
#9 0x08075960 in main ()

Seems to be wxBase problem - my version is 2.8.1.1-0ubuntu4. Hmmm... I will
install wxWidgets from source and recompile pgAdmin...

Regards,
Ivan

Ivan Pavlov wrote:
>
>
> On 5/18/07, *Dave Page* <dpage(at)postgresql(dot)org
> <mailto:dpage(at)postgresql(dot)org>> wrote:
>
> I haven't seen a pgadmin logfile from you yet so I wouldn't know what is
> new and what isn't!
>
>
> What I meant is - all entries in the log are timestamped and there were
> only some month old sql errors in there.

Ahh, right.

> They don't help much unfortuantely. Can you fire it up in gdb and get a
> plain old back trace please?
>
>
> Here it is:
> #0 0xb77225ae in wxStringBase::find () from /usr/lib/libwx_baseu-2.8.so.0
> #1 0xb772534a in wxStringBase::find () from /usr/lib/libwx_baseu-2.8.so.0
> #2 0xb77253e4 in wxString::Find () from /usr/lib/libwx_baseu-2.8.so.0
> #3 0x08347dee in isPgApp ()
> #4 0x0807ace5 in pgAdmin3::InitPaths ()
> #5 0x0808168e in pgAdmin3::OnInit ()
> #6 0x08084d01 in wxAppConsole::CallOnInit ()
> #7 0xb7700250 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
> #8 0xb7700327 in wxEntry () from /usr/lib/libwx_baseu-2.8.so.0
> #9 0x08075960 in main ()
>
> Seems to be wxBase problem - my version is 2.8.1.1-0ubuntu4 . Hmmm... I
> will install wxWidgets from source and recompile pgAdmin...

I just dusted off my FC6 VM, installed PostgreSQL 8.2.4 and wxGTK 2.8.4,
both from source and SVN-trunk builds and runs perfectly (all bar a
minor toolbar button sizing issue that you wouldn't notice unless you
stared at pgAdmin as much as I do!!).

The pgAdmin function it's dying in is this:

bool isPgApp(const wxString &app)
{
if (!wxFile::Exists(app))
return false;

wxArrayString output;

wxExecute(app + wxT(" --version"), output, 0);

if (output[0].Contains(wxT("PostgreSQL")))
return true;

return false;
}

Given that it appears to be bailing in wxStringBase::find, I'd guess
that's the "if (output[0].Contains(wxT("PostgreSQL")))" line. There's
nothing wrong with that, so I'd have to guess there's a problem with
your wx build.

I agree with your idea - grab wxGTK 2.8.4 and try rebuilding it from
scratch (there's a script in xtra/wx-build in the pgAdmin source tree to
build all the different combinations of build that we support).

Regards Dave