Re: inconsistent validation results

On Tue, 15 Apr 2008 22:42:42 +0530
Julian Abbs wrote:

> Hi,
> i have been validating a number of pages for accessibilty testing. At
> one point i accidentally validated the same page twice (by direct file
> upload) not a serious problem except that i noticed that i got a
> different number of errors on the two reports (93 the first time then
> 102) despite the fact that i used exactly the same file, in fact
> repeatedly validating the same page continues to give a different
> number of errors though i do at least get one or the other number
> every time. I note that the source shown in "Below is the source input
> I used for this validation:" differs between the two reports. Since
> then i have re-validated all the pages which previously passed
> validation and am now finding that every page is reporting a large
> number of errors related to non-closure of meta tags. When previously
> validated the source in "Below is the source input I used for this
> validation:" shows no unclosed tags, the actual source does in fact
> have a number of unclosed tags. I'm extremely concerned at this
> inconsistency as it suggests that the validator is not validating the
> exact source as uploaded and in fact cannot be relied upon without
> making a line by line comparison of the source input used for
> validation and the actual source. At this point i'm unable to provide
> the exact source used due to client confidentiality but may be able to
> do so if necessary. I have not previously noted any serious
> inconsistencies except between different validation methods (e.g. file
> upload versus validate by URL or cut and paste) but of course i do not
> usually validate a page more than once. I'm sending this email prior
> to submitting a bug report via bugzilla please can you assist.

A. The validator was improved a couple of months back, it now catches
more bad coding.

B. Probably not worth even looking at a bug report until you can reduce
the number of errors in your code, to specify where the differences are,
unles you can create test cases with minimum errors.


All shall be well, and all shall be well, and all manner of things shall
be well

 - Julian of Norwich 1342 - 1416

Received on Tuesday, 15 April 2008 18:25:06 UTC