> (2) Have everything available in a set of parallel
> accept-silver-${qetest.processor} directories and then define a
> qetest.processor=[XalanC|XalanJ-I|XalanJ-C] (or something similar)
property
> in build.xml or test.properties. The accept-silver-${qetest.processor}
> directories would contain the silver ("not quite gold") files for each
> specific type of processor.
It's not totally clear if these would be a complete reproduction of the
so-called gold files, or just the deltas (maybe I didn't read carefully
enough... from paragraph 2 I think you mean just the deltas). If a
complete reproduction, the I give a solid -1. If just the deltas, then I
give a +0, I guess.
I would feel better if we:
1) Enumerated the differences that need to be processor dependent.
2) See how many of these differences could be handled via better
comparators, probably with configuration options.
3) Possibly for those cases that can't be handled by (2), use a file naming
convention, like axes01-xalanc.xml.
I hate to see more trees added to an already complex testing
infrastructure.
-scott
[EMAIL PROTECTED] wrote on 11/28/2002 05:52:36 PM:
> Hello all,
>
> We are in the process of fascilitating the testing of XSLTC with the
> existing Xalan test suite. Some bugzilla bugs are open and being
addressed
> for issues with our test harness -- 13384 regards the tests for which
XSLTC
> generates the correct output but 'fails' as the output does not match the
> correct output from Xalan. While some effort has been made to re-write
test
> cases to remain processor independent, as well as improve the
intelligence
> of file comparators, it's not possible in every case. The following
> solution was proposed:
>
> We can setup an 'accept' tree, parallel to conf, in the tests tree. The
> subdirectories of accept would be output, idkey, intl, processorinfo, and
> maybe others like sort. These subdirctories would contain tests which
have
> different yet correct outputs from the different processors.The smoketest
> targets can be expanded to also run tests from accept. The remaining
> unanswered question is how to support the multiple 'gold' files. Possible
> options that were proposed for supporting multiple gold files are as
> follows:
>
> (1) A CVS-based solution. Having multiple copies in CVS of the
accept-gold
> directory with the same name; one for each of the the different
processors.
> To test a certain processor we could checkout the version of the
> accept-gold directory tagged for that processor.
>
> (2) Have everything available in a set of parallel
> accept-silver-${qetest.processor} directories and then define a
> qetest.processor=[XalanC|XalanJ-I|XalanJ-C] (or something similar)
property
> in build.xml or test.properties. The accept-silver-${qetest.processor}
> directories would contain the silver ("not quite gold") files for each
> specific type of processor.
>
> Some additional issues to note: In addition to dividing by processor, in
> the future the Xalan part may also be divided by setting of
output-escaping
> options, like which sets of named entities will be used and whether hrefs
> in HTML will have the % escaping.
>
> Comments? Suggestions? Opinions? Please give your input and vote by
> December 2nd. Thanks!
>
> Cheers,
> Gordon
>
> ----
> Gordon Chiu
> [EMAIL PROTECTED]
>
