[Catalyst] Catalyst::Plugin::Server::XMLRPC prefix option broken under lighttpd
Hi,
Catalyst versions affected: 5.69/5.7
Server: lighttpd 1.4.11_1
For the test case assume the following:
package MyApp;
__PACKAGE__-config(
xmlrpc = {
path = 'xmlrpc',
prefix = 'xmlrpc',
show_errors = 1
}
);
...
package MyApp::Controller::XMLRPC::Session;
sub authenticate : XMLRPC {
...
}
When using myapp_server.pl script, the entry is set to
http://localhost:3000/xmlrpc, so when call for 'session.authenticate', it
will dispatch properly to XMLRPC::Session::authenticate sub.
However once the code executes under fastcgi and lighttpd, the dispatch only
works when call is done with xmlrpc prefix, such
as 'xmlrpc.session.authenticate', when the call is done
as 'session.authenticate', it complains about not finding the method that can
handle the request. This should work however with the correct prefix being
set.
I noticed in Catalyst::Plugin::Server::XMLRPC::_deserialize_xml that forward
path is being set correctly at $self-forward_path( $fwd_path ), with the
specified prefix, so I am wondering if something else outside the plugin is
causing the problem that's related to fcgi environment.
Thanks.
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
[Catalyst] Weird variable chop out of manually set response body
I ran into a weird thing today (coming back to Catalyst after a long
hiatus so I hope I'm not just being obtuse; I love all the changes
I've found so far). Catalyst 5.7000, perl 5.8.6. Using test server.
This in MyApp::Controller::Word:
sub single : Path : Args(1) {
my ( $self, $c, $word ) = @_;
$c-response-body(Word: $word);
}
Called with http://host.local:3000/word/asdf
Gives: Word:
But modifying this line with a newline:
$c-response-body(Word: $word\n);
Gives the correct response: Word: asdf
So what gives? Expected behavior? If so, why? It didn't matter what
the variable was either. Things like $c-response-body(Word: .
$c) would not make it to output either.
-Ashley
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
[Catalyst] strange behaviour when controllers are being inherited
Hi, I just found something odd when trying to inherit from a base controller. I attached the complete application and debug output. From the debug output you can clearly see that MyApp powered by Catalyst as well as other things appear 4 times in the output and are being duplicated. This caused some concern. Please note that this only appears when running under fastcgi environment and lighttpd, I haven't tested the apache and standalone server does not seem to be affected. My layout is as follows: MyApp::Controller::Root MyApp::Controller::Welcome MyApp::Controller::Welcome::User ( inherits from MyApp::Controller::Welcome ) Versions of catalyst that I found affected: 5.69/5.7 Thank you. MyApp.tar.gz Description: application/tgz [debug] Debug messages enabled [debug] Loaded plugins: .. | Catalyst::Plugin::ConfigLoader 0.11 | | Catalyst::Plugin::Static::Simple 0.14 | '' [debug] Loaded dispatcher Catalyst::Dispatcher [debug] Loaded engine Catalyst::Engine::FastCGI [debug] Found home /usr/home/alex/svn/node3.grid0/sites/myapp.com/MyApp [debug] Loaded Config /usr/home/alex/svn/node3.grid0/sites/myapp.com/MyApp/myapp.yml [debug] Loaded components: .-+--. | Class | Type | +-+--+ | MyApp::Controller::Root | instance | | MyApp::Controller::Welcome | instance | | MyApp::Controller::Welcome::User| instance | '-+--' [debug] Loaded Private actions: .--+--+--. | Private | Class| Method | +--+--+--+ | /default | MyApp::Controller::Root | default | | /end | MyApp::Controller::Root | end | | /welcome/default | MyApp::Controller::Welcome | default | | /welcome/user/defau- | MyApp::Controller::Welcome::User | default | | lt | | | '--+--+--' [info] MyApp powered by Catalyst 5.7000 [debug] Debug messages enabled [debug] Loaded plugins: .. | Catalyst::Plugin::ConfigLoader 0.11 | | Catalyst::Plugin::Static::Simple 0.14 | '' [debug] Loaded dispatcher Catalyst::Dispatcher [debug] Loaded engine Catalyst::Engine::FastCGI [debug] Found home /usr/home/alex/svn/node3.grid0/sites/myapp.com/MyApp [debug] Loaded Config /usr/home/alex/svn/node3.grid0/sites/myapp.com/MyApp/myapp.yml [debug] Loaded components: .-+--. | Class | Type | +-+--+ | MyApp::Controller::Root | instance | | MyApp::Controller::Welcome | instance | | MyApp::Controller::Welcome::User| instance | '-+--' [debug] Loaded Private actions: .--+--+--. | Private | Class| Method | +--+--+--+ | /default | MyApp::Controller::Root | default | | /end | MyApp::Controller::Root | end | | /welcome/default | MyApp::Controller::Welcome | default | | /welcome/user/defau- | MyApp::Controller::Welcome::User | default | | lt | | | '--+--+--' [info] MyApp powered by Catalyst 5.7000 [debug] Debug messages enabled [debug] Loaded plugins: .. | Catalyst::Plugin::ConfigLoader 0.11 | | Catalyst::Plugin::Static::Simple 0.14
Re: [Catalyst] strange behaviour when controllers are being inherited
No, Just FCGI-0.67 Thanks. On Wednesday 19 July 2006 17:04, Matt S Trout wrote: Alex Pavlovic wrote: Hi, I just found something odd when trying to inherit from a base controller. I attached the complete application and debug output. From the debug output you can clearly see that MyApp powered by Catalyst as well as other things appear 4 times in the output and are being duplicated. This caused some concern. Please note that this only appears when running under fastcgi environment and lighttpd, I haven't tested the apache and standalone server does not seem to be affected. Are you using the FGCI process manager? ___ List: Catalyst@lists.rawmode.org Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/ Dev site: http://dev.catalyst.perl.org/
Re: [Catalyst] strange behaviour when controllers are being inherited
Alex Pavlovic wrote: No, Just FCGI-0.67 Thanks. On Wednesday 19 July 2006 17:04, Matt S Trout wrote: Alex Pavlovic wrote: Hi, I just found something odd when trying to inherit from a base controller. I attached the complete application and debug output. From the debug output you can clearly see that MyApp powered by Catalyst as well as other things appear 4 times in the output and are being duplicated. This caused some concern. Please note that this only appears when running under fastcgi environment and lighttpd, I haven't tested the apache and standalone server does not seem to be affected. Are you using the FGCI process manager? Try adding warn 'Starting up process id '.$$; immediately before __PACKAGE__-setup and see if it's getting loaded multiple times in different processes -- Matt S Trout Offering custom development, consultancy and support Technical Directorcontracts for Catalyst, DBIx::Class and BAST. Contact Shadowcat Systems Ltd. mst (at) shadowcatsystems.co.uk for more information + Help us build a better perl ORM: http://dbix-class.shadowcatsystems.co.uk/ + ___ List: Catalyst@lists.rawmode.org Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/ Dev site: http://dev.catalyst.perl.org/
Re: [Catalyst] strange behaviour when controllers are being inherited
Hi, Just tried it with FCGI::ProcManager 0.17 in standalone server mode. The problem goes away. Debug screen looks normal. I guess for some reason this only happens when running in static mode with builtin lighttpd proc manager. Thanks. On Wednesday 19 July 2006 17:04, Matt S Trout wrote: Alex Pavlovic wrote: Hi, I just found something odd when trying to inherit from a base controller. I attached the complete application and debug output. From the debug output you can clearly see that MyApp powered by Catalyst as well as other things appear 4 times in the output and are being duplicated. This caused some concern. Please note that this only appears when running under fastcgi environment and lighttpd, I haven't tested the apache and standalone server does not seem to be affected. Are you using the FGCI process manager? ___ List: Catalyst@lists.rawmode.org Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/ Dev site: http://dev.catalyst.perl.org/
Re: [Catalyst] strange behaviour when controllers are being inherited
Starting up process id:23065 at lib/MyApp.pm line 33. Starting up process id:23063 at lib/MyApp.pm line 33. Starting up process id:23066 at lib/MyApp.pm line 33. Starting up process id:23064 at lib/MyApp.pm line 33. On Wednesday 19 July 2006 17:37, Matt S Trout wrote: warn 'Starting up process id '.$$; ___ List: Catalyst@lists.rawmode.org Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/ Dev site: http://dev.catalyst.perl.org/
Re: [Catalyst] Weird variable chop out of manually set response body
I tried this on my setup but didn't get the chop you've reported. I've
attached my controller for you to look at and see if there are any other
differences.
I'm running this on a unix box that is fairly up to date, Cat 5.7 etc. What's
your underlying OS?
If you recently upgraded an application you might need to rebuild the stuff in
scripts with 'catalyst.pl -force -scripts [your app]'
I'd try updating your local perl a bit with install Task::Catalyst and so forth
to make sure you have all the newer versions. Or you can run a script to force
CPAN to update everything out of date on your local box, which has solved
problems for me in the past (although could introduce other problems).
Try the attached controller and see if it still have the mentioned trouble. If
we can ID your exact setup maybe we can find someone on the list that matches
it to try the controller as well.
let me know what you find out! --john
- Original Message
From: apv [EMAIL PROTECTED]
To: The framework catalyst@lists.rawmode.org
Sent: Thursday, July 20, 2006 3:52:20 AM
Subject: [Catalyst] Weird variable chop out of manually set response body
I ran into a weird thing today (coming back to Catalyst after a long
hiatus so I hope I'm not just being obtuse; I love all the changes
I've found so far). Catalyst 5.7000, perl 5.8.6. Using test server.
This in MyApp::Controller::Word:
sub single : Path : Args(1) {
my ( $self, $c, $word ) = @_;
$c-response-body(Word: $word);
}
Called with http://host.local:3000/word/asdf
Gives: Word:
But modifying this line with a newline:
$c-response-body(Word: $word\n);
Gives the correct response: Word: asdf
So what gives? Expected behavior? If so, why? It didn't matter what
the variable was either. Things like $c-response-body(Word: .
$c) would not make it to output either.
-Ashley
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
Word.pm
Description: Binary data
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
Re: [Catalyst] Catalyst::Plugin::Server::XMLRPC prefix option broken under lighttpd
Hi,
Just an update on the situation. The prefix is broken only when your entry
point ( path ) and prefix are the same. As shown in example sent earlier.
This works under standalone myapp_server.pl but not under fastcgi because CGI
engine sets the base path to $ENV{SCRIPT_NAME} in prepare_path, this in
turn causes $location in Catalyst::Request::path to be set to this, well
you guess what happens afterwards $path =~ s/^(\Q$location\E)?//;, a little
bit of substitution which takes the prefix out, and causes xmlrpc method not
to be found.
So to cut the story short, don't use the same prefix and entry point name like
I did and you will save yourself time debugging.
Thanks.
On Wednesday 19 July 2006 00:12, Alex Pavlovic wrote:
Hi,
Catalyst versions affected: 5.69/5.7
Server: lighttpd 1.4.11_1
For the test case assume the following:
package MyApp;
__PACKAGE__-config(
xmlrpc = {
path = 'xmlrpc',
prefix = 'xmlrpc',
show_errors = 1
}
);
...
package MyApp::Controller::XMLRPC::Session;
sub authenticate : XMLRPC {
...
}
When using myapp_server.pl script, the entry is set to
http://localhost:3000/xmlrpc, so when call for 'session.authenticate', it
will dispatch properly to XMLRPC::Session::authenticate sub.
However once the code executes under fastcgi and lighttpd, the dispatch
only works when call is done with xmlrpc prefix, such
as 'xmlrpc.session.authenticate', when the call is done
as 'session.authenticate', it complains about not finding the method that
can handle the request. This should work however with the correct prefix
being set.
I noticed in Catalyst::Plugin::Server::XMLRPC::_deserialize_xml that
forward path is being set correctly at $self-forward_path( $fwd_path ),
with the specified prefix, so I am wondering if something else outside the
plugin is causing the problem that's related to fcgi environment.
Thanks.
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
___
List: Catalyst@lists.rawmode.org
Listinfo: http://lists.rawmode.org/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/catalyst@lists.rawmode.org/
Dev site: http://dev.catalyst.perl.org/
