[rfc] make test ignored or valid only for specific JDK

[Jonathan Gibbons]

Jiri,

I presume you are talking about a set of tests that is distinct from any 
of the jtreg test suites in an OpenJDK forest, such as langtools/test, 
jdk/test, etc.   Those test suites are tied to the code in the overall 
forest, and so are inherently versioned without any additional notation.

There are a variety of existing mechanisms you could use to address this 
problem, without adding any new features.

1. Use test lists.   You can keep lists identifying which tests are 
valid in each version, and then specify files containing those lists on 
the command line with the "@file" mechanism

2. Use exclude lists.  You can keep lists identifying which tests are 
/not/ valid in each version, and then specify files containing those 
lists on the command line with the "-exclude file" mechanism

3. Keywords. You can tag individual tests with any keywords you choose 
to use, as long as you also declare the overall set of valid keywords in 
the TEST.ROOT file.   You can filter the set of tests to be executed 
with the "-k expr" command line option.

4. @requires. This is a relatively new feature. You can put a 
declarative tag in a test description of the form "@requires expr" where 
there is a variety of terminals, of which jdk.version is but one.   So 
you could add a line like
     @requires  jdk.version <=1.6 || jdk.version >1.7 || jdk.version = 
"1.4.2"
Not as compact as your proposal, but it exists today.

-- Jon

On 03/26/2015 12:46 AM, Jiri Vanek wrote:
> Hello!
>
> I'm maintaining few thousand tests for various jdks (generally jdk 
> 1.5-jdk 9) . However, current workarounds about  which test is valid 
> for which jdk are insufficient.
>
> I came to conclusion, that some kind of @validfor-jdk-version or 
> @validfor or @ignoredfor or enhancing of @ignore would solve my issues.
>
> Currentl I'm inklining to @validfor  or @validfor-version as 
> (currently (with this 7years history)) it is not necessary to 
> distinguish anything else then java.version. My suggested format will 
> be white-chars separated list of conditions with control words of = < 
> > <= >= followed  by version string and with evaluation based on 
> version sort and on or between predicates
>
> So eg:
> @validfor  <=1.6 >1.7 =1.4.2
>
> If java.version  will not be  valid for any of predicate then the test 
> will be handled in same way as if @ignore is  specified. If it is 
> valid, or annotation missing at all, test will be normally proceed.
>
> Do you think enhancement like this is acceptable?
>
> If so ,I will prepare webrew soon, and will post it for review.
>
> Thank you!
>
>  J.