RFR: CODETOOLS-7903748 - jcstress: Test list should honor concurrency settings [v4]

[Jiří Vaněk]

On Wed, 3 Jul 2024 19:21:08 GMT, PM <duke at openjdk.org> wrote:

>> Jiří Vaněk has updated the pull request incrementally with two additional commits since the last revision:
>> 
>>  - Moved listing of tests to separate method: listTests
>>  - -l description moved to constant
>
> jcstress-core/src/main/java/org/openjdk/jcstress/infra/runners/TestConfig.java line 264:
> 
>> 262:                 ", strideCount: " + strideCount +
>> 263:                 ", cpuMap: " + cpuMap +
>> 264:                 ", " + jvmArgs + "}";
> 
> kindly modify this to return a valid json, perhaps try running this method locally once in your machine, and copy paste that json in any online json formatter to check if you encounter any errors

I never had an intention to return valid json.  It is used only in verbose output, to differentiate individual variants of tests.  Wihtout the test id, it do no have sense, and the extended info in {} is... complicated. See eg:

org.openjdk.jcstress.samples.api.API_01_Simple {[actor1, actor2], spinLoopStyle: Thread.onSpinWait(), threads: 2, forkId: 0, maxFootprintMB: 64, compileMode: 0, shClass: (PG 0, CG 0), (PG 0, CG 0), strideSize: 256, strideCount: 40, cpuMap: null, [-XX:+UseBiasedLocking]}
org.openjdk.jcstress.samples.api.API_01_Simple {[actor1, actor2], spinLoopStyle: Thread.onSpinWait(), threads: 2, forkId: 0, maxFootprintMB: 64, compileMode: 0, shClass: (PG 0, CG 0), (PG 0, CG 0), strideSize: 256, strideCount: 40, cpuMap: null, [-XX:-UseBiasedLocking]}



To highlight it is not json, was the original usage of `=`,  but I did not had super strong opinion.
As the output should remain mainly human readable, I'm quite against full json (thus bracket before name, and id before actorNames and jvmArgs).  But on contrary,  the list is indeed huge.

The main motivation is the line of ""All matching tests combinations - " + testsToPrint.size()" thus real number of run tests. The listing itself is (very) useful candy on top:

java -jar tests-all/target/jcstress.jar -v -l  -t org.openjdk.jcstress.samples.api.API_01_Simple 
...
All matching tests combinations - 96
...

x

java -jar tests-all/target/jcstress.jar -l  -t org.openjdk.jcstress.samples.api.API_01_Simple 
...
All matching tests - 1
...



So back to verbose listig output format - I intentionally kept it semi parseabel - grep/sed  on output, and readable enough: name  info in {} , all firelds with key id, and obvious arrays in [] without key.
AFAICJ, the full json will lower the human readabiity. 
What about this - with "-v" it would print it as it is - human readable, fully distinguishible.  on "-vv or -vvv"  it would print complete fully qulified json. WDYT?

-------------

PR Review Comment: https://git.openjdk.org/jcstress/pull/149#discussion_r1665173411