- From: Terje Bless <link@pobox.com>
- Date: Fri, 27 Jul 2001 21:02:53 +0200
- To: W3C Validator <www-validator@w3.org>
- cc: Robert.W.Blackburn@apcc.com
BTW, you need to send messages to <www-validator@w3.org> and not <www-validator-request@w3.org>. If you haven't subscribed to the list, messages will be bounced to a moderator and this detour may have introduced the wrong address somewhere along the way. On 27.07.01 at 11:11, Robert.W.Blackburn@apcc.com wrote: >Q: Does running my version of the validator against a known valid > page give the same error >A: Using http://validator.w3.org/ does seem to work. Ok, so we know it's caused by our error reporting code. >Q: What information does running NSGMLS by hand give >A: I ran NSGMLS by hand and attached both the HTML page and the error > text (See attached file: foo.html)(See attached file: errors.txt) Ok, so we know the parser is working and giving normal error messages. >Q: you did get the binary Iconv distribution from ActiveState and > installed it? >A: Yes I did, and it seemed to install correctly. Then it likely did or you would have gotten an error from PPM. >Q: The version of the "check" CGI application >A: $VERSION = q$Revision: 1.122 $; Ok. I'm guessing it's a problem with our error parsing code. Could you try the attached version? I've modified the path to nsgmls.exe to use UNIXy paths -- flop the slashes back the wrong way if your Windows variant don't understand those :-) -- eliminated some code, and modified error parsing. I'm guessing the problem stems from either the backslashes in the path (which Perl thinks are escape characters) or from that extra ":" that gets introduced as part of "D:\" (we use the colons in nsgmls.exe's output as position markers to extract the info we need). The attached version /should/ fix it if that's the problem. One indication of this would be if the error message you got was different each time. Apart from the "Uh Oh! I got the following unknown error" spiel; did the message afterwards -- the "invalid comment declaration" bit -- change depending on which page you tried to validate?
Received on Friday, 27 July 2001 15:03:00 UTC