[CI] Fix CI script by using absolute path to ROSENBR.SIF

[jfowkes]

No description provided.

[jfowkes]

@nimgould thank you very much for the fixes, the latest change to the sifdecoder bash script seems to require that absolute paths to SIF files are passed to sifdecoder as is the case for sifdecoder_standalone. Is this intentional? If so this PR fixes the CI tests.

[nimgould]

Yes, this is intentional, we say that the SIF file should either be in the current directory, or the SIF repository (aka MASTSIF). The test options in the makemaster version change directory to $(SIFDECODE)/sif before running tests, I presume that meson could copy (or link) the SIF files if this is not possible?

[nimgould]

I also notice that there were many lines saying

tput: No value for $TERM and no -T specified

in the actions test output. I have no idea where this comes from
or what it means!

[jfowkes]

Excellent thank you, yes I can change the CI tests to cd into $(SIFDECODE)/sif, will do that now.

Regarding the "no value for $TERM" I have already fixed this by setting TERM=xterm. This is a strange CI issue.

[nimgould]

Perfect, let's see if the tests now complete

[jfowkes]

I should clarify that what seems to have broken is that the sifdecoder bash script is no longer looking for SIF files in $MASTSIF, the CI scripts currently set MASTSIF=$(SIFDECODE)/sif but this is no longer being picked up.

[nimgould]

Is $(SIFDECODE) being set?

[jfowkes]

Yes the CI scripts currently set:

env:
    ARCHDEFS: /home/runner/work/SIFDecode/SIFDecode/ARCHDefs
    SIFDECODE: /home/runner/work/SIFDecode/SIFDecode
    MASTSIF: /home/runner/work/SIFDecode/SIFDecode/sif

which are all the correct paths to where the CI installs things.

[nimgould]

Well that is mighty odd, but it also seems to happen to me locally ...

% sifdecoder ROSENBR

ERROR: file ROSENBR.SIF is not known in directories
/tmp or $MASTSIF

% ls $MASTSIF/ROSENBR.SIF
/home/nimg/problems/sif/ROSENBR.SIF

I will put on my cape and deerstalker, and examine with a large magnifying glass ...

[jfowkes]

Thank you, also any chance we could add a WARNING: in front of the new Failed to set HNS to 4 -- skipping line in sifdecoder_standalone to keep the old bash script behaviour? Makes parsing parameter failures easier.

[nimgould]

OK, fixed, done and pushed

[jfowkes]

Wonderful thank you, closing this pull request as it is no longer required.