run_tests "ERROR: Cannot open connected subprocess" for several tests

[astroJarHead]

After installing v 2.17 the result of test/run_tests was:

Test summary:      15 failed
    111 passed
      2 skipped
      1 xfailed

All the error message contained text similar to the following except the sub-process called was different:

Output
======
cl> copy noao$artdata/numrecipes.x .
cl> copy noao$artdata/gammln.c .
cl> softools
cl> xc -I$iraf/include test-gasdev.x numrecipes.x gammln.c
cl> task $test_gasdev = test-gasdev.e
cl> test_gasdev
ERROR: Cannot open connected subprocess (test-gasdev.e)
     called as: `cl ()'
Error while reading login.cl file - may need to rebuild with mkiraf
Fatal startup error.  CL dies.

Attached is the output of run_tests
out.run_tests.txt

sysgen did seem to finish without problems prior to the tests running. At least
that I could note with my limited experience. The output from sysgen completed with:

Updated 31 files in libpkg.a
xc -c -w -p noao x_quad.x
xc -Nz -p noao x_quad.o libpkg.a -lxtools -o xx_quad.e
move `xx_quad.e' to `noaobin$x_quad.e'
purge directory `noaobin$'
!date
Wed Dec 14 09:30:16 MST 2022

[olebole]

It would still be useful to attach the build log.
The reason here is unclear; it seems that with the system you (sometimes) can't compile own tasks -- some tasks however seem to work.

I would recommend to try them out manually in the "ecl" shell; first create a file test-gasdev.x with the content from here. Then start ecl and run the commands as in the log (resp. the link above). The log shows (lines starting with cl> ) what you should input and what you should expect. The xc task you have shown above should produce an output file test-gasdev.e, which should have the executable flag set.

I am afraid that you need to do some debugging here yourself.

[olebole]

The build log look OK.

[astroJarHead]

👍

[astroJarHead]

The manual test using the "ecl" shell was successful as the newly defined task test_gasdev did run. There was a difference in numerical precision:

ecl> copy noao$artdata/numrecipes.x .
ecl> copy noao$artdata/gammln.c .
ecl> softools
generic lroff mkapropos mkpkg mkttydata rmfiles wtar xyacc
hdbexamine memchk mkhelpdb mktags rmbin rtar xc

softools> xc -I$iraf/include test-gasdev.x numrecipes.x gammln.c
softools> task $test_gasdev = test-gasdev.e
softools> test_gasdev
Mean = 0.0000601690205889
Variance = 1.0002424406825960
Skew = 0.0001202033929215
Kurtosis = 3.0006296332838071

I notice that the prompt in the the output I recorded from run_test that began this discussion indicated the "cl" shell
was in use. I did repeat the manual test with the "cl" shell and it ran successfully.

[olebole]

OK; so everything seems to work now. And you can start IRAF and do something with it. Right?

[astroJarHead]

Yes. Other than the oddity with the run_tests all seems well. I appreciate your assistance.