remove non piraha parser from Flesh

[Roland Haas]

Similarly to #2452 In the Turing release announcements: http://einsteintoolkit.org/about/releases/ET_2020_05_announcement.html we are deprecating the non-Piraha parser which increasingly no longer parses all existing parfiles (eg the non-piraha parfile parser does not handle $foo = 42 types of variable assignments).

It should thus be removed from the code.

[Roland Haas]

Removing the parser code would resolve https://bitbucket.org/cactuscode/cactus/pull-requests/103/cactus-avoid-comparing-signed-and-unsigned

[Roland Haas]

This could still be done and is only code removal from src/util/ParseFile.c

[Roland Haas]

Has there been any progress removing this?

[Roland Haas]

Has there been any progress removing this?

[Roland Haas]

This would be suitable for a medium hackathon task.

[William Gabella]

Took this ticket for the Hack-a-thon, 2 Feb 2022. Removing all the local routines, etc, related to the “old way” of parsing the file, and of course keeping the “new way,” with cctk_PirahaParser() function. This is in Cactus/repos/flesh/src/util/ParseFile.c , as Roland mentions above.

Re-compiling with (from the Cactus folder)

./simfactory/bin/sim build -j2 --thornlist ../einsteintoolkit.th

and watching for the “COMPILING … ParseFile.c” to go by.

And running HelloWorld first, as

./simfactory/bin/sim create-run helloworld \
    --parfile arrangements/CactusExamples/HelloWorld/par/HelloWorld.par

If that is not broke, then finally the testsuite with

make sim-testsuite

Format for the teststuite is “make <config name>-testsuite. It gave a set of defaults and I chose them all; testing exe/cactus_sim .

[William Gabella]

Roland points out the next level is to look for functions in the deleted parts of ParseFile.c that are called from external sources and are ONLY used in the old part of ParseFile.c.

[William Gabella]

Seems to only affect the ParseFile.c . No other changes.

[William Gabella]

Passed all tests in make sim-testsuite:

 Summary for configuration sim

    Time                     -> Wed Feb  2 05:47:18 PM CST 2022
    Host                     -> fedora
    Processes                -> 2
    User                     -> gabella

    Total available tests    -> 346
    Unrunnable tests         -> 14
    Runnable tests           -> 332
    Total number of thorns   -> 258
    Number of tested thorns  -> 99
    Number of tests passed   -> 332
    Number passed only to
               set tolerance -> 195
    Number failed            -> 0

‌

[Roland Haas]

Is there a pull request to approve?

[William Gabella]

Created the gabella/piraha branch via BitBucket, re-installed the master on my laptop, checked out the gabella/piraha branch putting in the edited ParseFile.c ---with the removed non-Piraha code---into the directory Cactus/repos/flesh/src/util . For the git push, required I set the upstream
git push --set-upstream origin gabella/piraha
and logged in with username and password, despite having SSH keys set. Also saw a warning about username logins being disabled and app passwords being necessary, oh, for HTTPS logins. Should have used the SSH address for the repos.

[William Gabella]

Also re-compiled and ran the testsuites
make sim-testsuite
the first time running it was very slow and essentially stopped making much progress at GiRaFFE_tests_ThreeWave (from GiRaFFE), I think. Stopped and re-ran and that went smoothly. The results are

Summary for configuration sim

    Time                     -> Thu Feb 17 06:38:31 PM CST 2022
    Host                     -> fedora
    Processes                -> 2
    User                     -> gabella

    Total available tests    -> 346
    Unrunnable tests         -> 14
    Runnable tests           -> 332
    Total number of thorns   -> 258
    Number of tested thorns  -> 99
    Number of tests passed   -> 330
    Number passed only to
               set tolerance -> 193
    Number failed            -> 2

  Tests passed:

    tov (from ADMMass)
    tov_carpet (from ADMMass)
...
    teukolskyParity (from WeylScal4)
    magnetized_explosionTOV (from particle_tracerET)
    magnetized_explosionTOV (from smallbPoynET)

  Tests failed:

    Kerr-Cartoon (from AHFinderDirect)
    Kerr-Cartoon-EE (from AHFinderDirect)

‌

Curiously the previous tests had no failures, 2 Feb 2022, posted above. The reports are

AHFinderDirect: Kerr-Cartoon
Failure: Cactus exited with error code 1.
AHFinderDirect: Kerr-Cartoon-EE
Failure: Cactus exited with error code 17.

‌

[William Gabella]

Please review.

[Roland Haas]

I will be busy with proposal this weekend. Will try to review next week. Not sure what would make the AHFinderDirect tests fail. You can find the actual log output in TEST/sim/AHFinderDirect/Kerr-Cartoon.log (or so) which may, or may not, contain useful information.

[Roland Haas]

Pull request is here: https://bitbucket.org/cactuscode/cactus/pull-requests/139

[Roland Haas]

Please apply.

[Roland Haas]

Applied as git hash 2ba01b5b "Cactus: remove non-Piraha parser" of cactus