W3C home > Mailing lists > Public > public-qa-dev@w3.org > August 2007

Re: validator/httpd/cgi-bin check,1.556,1.557

From: olivier Thereaux <ot@w3.org>
Date: Thu, 23 Aug 2007 09:52:19 +0900
Message-Id: <3CEDEC32-E7E9-4AA1-BE64-368B1E713560@w3.org>
Cc: "QA-dev" <public-qa-dev@w3.org>
To: Ville Skyttä <ville.skytta@iki.fi>


On Aug 23, 2007, at 00:06 , Ville Skyttä wrote:

>
> Hi Olivier,
>
> On Wednesday 22 August 2007, you wrote:
>
>> * silencing some debug output
> [...]
>> @@ -2546,9 +2546,9 @@
>>    # beyond EOL.  If you see this warning in your web server logs,  
>> please
>>    # let the validator developers know, see
>> http://validator.w3.org/feedback.html if ((my $l =
>> length($File->{Content}->[$err->{line}-1])) < $err->{char}) { -
>> warn("Warning: reported error column larger than line length " . -
>> "($err->{char} > $l) in $File->{URI} line $err->{line}, " . -
>> "OpenSP bug? Resetting to line length.");
>> +    #warn("Warning: reported error column larger than line length  
>> " .
>> +    #     "($err->{char} > $l) in $File->{URI} line $err->{line},  
>> " .
>> +    #     "OpenSP bug? Resetting to line length.");
>>      $err->{char} = $l;
>>    }
>
> Do you actually see these messages in the v.w.o logs?

oh, yes...
# grep "reported error column larger than line length" error.log | wc -l
244019
and that's not even 4 days of logs, on one of the 3 servers.

Granted, every single case seems to yield at least a few dozen  
warning lines, and is generally revalidated a few times, so we don't  
have 200000 test cases, but we still have tons of them which we are  
not using, while the error logs for the validator are reaching  
gigabytes. I'm trying to bring that back to a more reasonable level.

>   This message was added IIRC on Terje's request hoping that a  
> reproducer will be found and can thus possibly be fixed in OpenSP.   
> Without the message, finding a reproducer will be much harder...

Understood, I should have made my homework and tried to remember the  
context of this code. The warn()ings are only commented out, and we  
can always bring them back on for a period of time. I think, however,  
that such debug output should not be on by default for a production  
instance.
Anyway, here's a test case that can be used (and which is not going  
to change):
http://qa-dev.w3.org/wmvs/HEAD/dev/tests/bogus-fpi3.html

-- 
olivier
Received on Thursday, 23 August 2007 00:51:25 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 19 August 2010 18:12:48 GMT