merge test_run and coverage_run into a single command in coverage script

[Vijay Kumar Banerjee]

The objectives of this ticket are TBD

[Chris Johns]

I am not sure this is possible because test_run is run once per test and coverage_run is run once at the end.

Is covoar going to change?

[Vijay Kumar Banerjee]

There was a suggestion to merge these two, but no clear objectives were decided, I think it's better to keep them separate.
Shall we close the ticket ?

This change was supposed to be in the script. There's another ticket filed for changes in covoar
please see the following ticket

​https://devel.rtems.org/ticket/3441

[Chris Johns]

Replying to thelunatic:

There was a suggestion to merge these two, but no clear objectives were decided, I think it's better to keep them separate.

I am not sure yet. It comes down to covoar and if it has to run as a single instance or on each test on common data.

Shall we close the ticket ?

Why?

This change was supposed to be in the script. There's another ticket filed for changes in covoar
please see the following ticket

​https://devel.rtems.org/ticket/3441

I do not understand how this relates.

[Vijay Kumar Banerjee]

Replying to Chris Johns:

Replying to thelunatic:

There was a suggestion to merge these two, but no clear objectives were decided, I think it's better to keep them separate.

I am not sure yet. It comes down to covoar and if it has to run as a single instance or on each test on common data.

I is needed to run them as a single instance? I mean, would there be any advantage of it?

Shall we close the ticket ?

Why?

This change was supposed to be in the script. There's another ticket filed for changes in covoar
please see the following ticket

​https://devel.rtems.org/ticket/3441

I do not understand how this relates.

[Chris Johns]

Replying to thelunatic:

Replying to Chris Johns:

Replying to thelunatic:

There was a suggestion to merge these two, but no clear objectives were decided, I think it's better to keep them separate.

I am not sure yet. It comes down to covoar and if it has to run as a single instance or on each test on common data.

I is needed to run them as a single instance? I mean, would there be any advantage of it?

This is what we need to examine which we can do there.

Covoar has evolved independently to rtems-test and as time has gone on we have settled on rtems-test and the standard interface for users to testing executables and covoar has been brought under it.

Covoar wants to create a union of all the instructions tested across all test executables. Running on each executable one at a time lets the overall task be split and run in parallel automatically as separate jobs which rtems-test does already. If we can process each test and generate the needed data we can look at making the final report at the end of the run.

I think we need to simplify covoar. I wonder if moving to the report generation would be better. We could use markdown or rest and we could create more complex reports via Python which is easier to do.

It would also let covoar concentrate on what it needs to do and that is the low level analysis of the trace data against the executable's image.