#2 split_message croaks on some OpenSP messages
validating http://www.xyzzy.claranet.de/home/test/res.html through SPO, and calling split_message() on one message received, SPO croaks with following message:
| Unexpected error message format ((invalid
location):47239440633056.2322:W: URL | Redirected to
| "http://www.xyzzy.claranet.de/w3c/html2070.dtd"
| ) at /usr/local/lib/perl/5.8.8/SGML/Parser/OpenSP.pm line 45
Which seems to point to the following regexp :
my @comp = split(/(?:^|\n)(?:<[^>]+>)?\Q$name\E:(\d+):(\d+):\s*/, $mess);
in SGML::Parser::OpenSP::split_message
I am not familiar enough with OpenSP's API to know whether the bug is in OpenSP's message not following its recommended formats, or if the regexp should be amended to be more flexible.
More info:
http://www.w3.org/Bugs/Public/show_bug.cgi?id=4516
http://lists.w3.org/Archives/Public/www-validator/2007Apr/0118.html
Logged In: YES
user_id=188003
Originator: NO
olivier, could you remind me, is this still the case with the cvs version? (It hasn't changed since the last release I believe). If, then I think it would be wise to treat it as S::P::O bug with OpenSP taking a long long time to get updated.
Logged In: YES
user_id=1223947
Originator: YES
I am in the process of re-creating the test case for this one, as unfortunately the one I used to have (online) has changed.
The way to reproduce is:
* use a document with a SYSTEM id or a PUBLIC id outside of the catalog (plus a SI)
* the SYSTEM id URI redirects to a 404
I've made a test case over at:
http://www.w3.org/Bugs/Public/attachment.cgi?id=557
the version currently installed on my computer:
# $Id: OpenSP.pm,v 1.34 2007/12/06 05:43:11 hoehrmann Exp $
does exhibit the error:
Unexpected error message format ((invalid location):27594284.2322:W: URL Redirected to "http://www.w3.org/glrgrb"
) at /Library/Perl/5.8.8/darwin-thread-multi-2level/SGML/Parser/OpenSP.pm line 45
I can install the cvs version of S::P::O and test if that helps.
Thanks
olivier.