[Catalyst] Testing server hangs with Safari and some others

[Jonathan Rockway]

Thanks for the info on the CGI.  Has a bug been filed for this yet, or 
should I report one?

I don't think your network issue is related to Safari.  Try using nc:

$ nc localhost 3000
GET /
< response from catalyst >

This cuts out HTTP/1.1 and any other "overhead" that a real web browser 
adds.  I bet, though, that this hangs too.  Try running under the profiler:

$ perl -d:DProf script/myapp_server.pl

If you manage to get good profile data, you should see one sub taking a 
lot of "real" time, and then you know where the bug is.

Unfortunately, DProf *loses data* if you kill the program with C-c, so 
this might not help detect the hang at all.  There are apparently other 
profilers, but none of those work with Catalyst:

$ perl -d:Profile script/blog_server.pl
Died at /usr/lib/perl5/site_perl/5.8/Template/Timer.pm line 63.
Compilation failed in require at 
/usr/lib/perl5/site_perl/5.8/Catalyst/View/TT.pm line 6.
BEGIN failed--compilation aborted at 
/usr/lib/perl5/site_perl/5.8/Catalyst/View/TT.pm line 6.
Compilation failed in require at (eval 
51)[/usr/lib/perl5/5.8/base.pm:82] line 3.
        ...propagated at /usr/lib/perl5/5.8/base.pm line 85.
BEGIN failed--compilation aborted at /cygdrive/c/Documents and 
Settings/jrockway/blog_software/scrip
t/../lib/Blog/View/HTML.pm line 4.
Compilation failed in require at (eval 
50)[/usr/lib/perl5/site_perl/5.8/Catalyst/Utils.pm:252] line
3.
Compilation failed in require at script/blog_server.pl line 52.

:)

DProf doesn't work for me either, because Math::Pari is very broken and 
DProf agitates its brokenness.  *sigh* again :)

(BTW, sorry my first few posts to the list are so cynical.  It's just 
that I spend a good precentage of my life with Catalyst and Perl so I've 
obviously had a chance to discover weird corner cases that cause 
annoyance.  Actually, none of the bugs are Catalyst bugs, they're TT or 
Crypt::OpenPGP/Crypt::Random/Math::Pari bugs.  Glad to know they're 
software bugs and not user error on my part, though. :)

Anyway, hope this helps somewhat. :)

Regards,
Jonathan Rockway

apv wrote:
> Hmmm. I'm not using CGI (it hangs on the most minimal 
> install/welcome-page) and I have the most recent regular TT2 release.
>
> I'm glad you mention it though b/c I did have the problem you list with 
> another TT2 form (not in Catalyst) and I found on antoher list that 
> naming the CGI instance or calling it with a namespace fixed the 
> hanging. [% USE CGI(foo) %] or [% USE mycgi = CGI(foo) %].
>
> http://www.template-toolkit.org/pipermail/templates/2002-May/003242.html
>
> I wrote a chat server about 4 years ago that had the same problem of 
> hanging on Safari but not on the others. I can't remember how I solved 
> it now but there was something different about the way Safari handled 
> the socket blocking on reads between headers and body, I think. I'll 
> try to dig up that code and compare it to the test server code.
>
> -Ashley
>   


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 368 bytes
Desc: OpenPGP digital signature
Url : http://lists.rawmode.org/pipermail/catalyst/attachments/20060726/55a21173/attachment-0001.pgp