doubts on t/steps/auto_pmc-01.t

[coke]

In this test file, we have two very long lists of PMCs (@dummy_options) that are almost but not quite identical. Having had to modify the core list of pmcs several times in the past few days, it'd be nice if this test didn't have to be massaged (at least quite so much) each time.

• should these test be being run twice? They seem to be identical except for the undocumented change in pmc listings.
• should these lists be identical? If so, let's eliminate one copy.
• otherwise, let's keep one list and change the one entry, avoiding the duplication, and document what the difference in the list is for.
• if the list stays as a literal list, please reformat it so it follows the line length convention. makes tracking diffs easier.
• do we need to store the literal list? can't we pull this from the MANIFEST or something? (I can't tell what this test is testing.)

Thanks.

[coke]

(reformatted description)

[jkeenan]

Coke:

Let me go back and see how config/auto/pmc.pm has evolved since the test was first written.

kid51

[jkeenan]

Replying to coke:

* if the list stays as a literal list, please reformat it so it follows the line length convention. makes tracking diffs easier.

Handled in  r35961. I'll work on the other issues as time permits. I hope we can clear up TT #219 soon, as that will facilitate my working on this ticket.

Thank you very much.
kid51

[jkeenan]

Replying to coke:

* should these test be being run twice? They seem to be identical except for the undocumented change in pmc listings. * should these lists be identical? If so, let's eliminate one copy. * otherwise, let's keep one list and change the one entry, avoiding the duplication, and document what the difference in the list is for.

I used coverage analysis to learn (or re-learn) what the point of the tests in t/steps/auto_pmc-01.t were. The analysis indicated that certain tests -- those that involved the second of the two lists Coke mentioned -- were used to test these two lines in config/auto/pmc.pm:

179:        next if $pmc =~ /^const/;
228:        next if $pmc_file =~ /^const/;

(Line numbers as of r35995; they have since changed.)

However, inspection of src/pmc/ indicated that we no longer have any .pmc files whose names begin with const. Discussion on #parrot with Coke and NotFound suggested that it would be okay to rip those lines out. Once they were deleted, I could eliminate the second list and associated tests in t/steps/auto_pmc-01.t. This was done in  r36000.

(I can't tell what this test is testing.)

Added explanatory comment in  r36001.

Thank you very much.
kid51

[jkeenan]

On IRC, Coke reported satisfaction with state of t/steps/auto_pmc-01.t. Closing ticket.

Thank you very much.
kid51