[uf-dev] $PRODID$ normalization [was: hCard test 30...]

[Ryan King]

On Jun 21, 2006, at 10:49 AM, Dan Connolly wrote:

> On Wed, 2006-06-21 at 12:37 -0500, brian suda wrote:
>> Any application that is using the Test Suite will have different  
>> PRODIDs
>> and different SOURCE values. So in the test suites we used $PRODID$
>> which is replaced by the string in the /bin/test-xlstproc.pl so  
>> you can
>> customize it for your applications output.
>>
>> We are DIFF'ing the output from xsltproc (or other applications)  
>> to the
>> .vcf files. If someone uses their own application it will have a
>> different string for the value of PRODID, so instead of hard-coding a
>> valued in the .vcf files and then during DIFF throwing errors, we  
>> used a
>> variable $PRODID$ which is swapped out so the diff doesn't fail.
>>
>> does that make more sense?
>
> Well, no. Why not just take the output of x2v and change
> the PRODID line, and then do the diff against the expected
> results? Why should test-xlstproc.pl change when x2v change?

Well, test-xsltproc is made specifically for testing x2v. This is a  
case where it'd probably be easier to have it in the same repository  
as x2v, rather than the tests repository.

> In Edinburgh, Ryan gave an answer that made sense to me at
> the time, but I either can't remember it or am no longer
> convinced by it. It had something to do with encouraging
> implementations to spit out some PRODID, and not just
> skipping the property altogether.

Yeah, that was why. It may not be worth the trouble.

-ryan

--
Ryan King
ryan at technorati.com