[bhuvan-somisetty]

Fixes #636

The windows MATLAB and Octave lines in mkconcore.py now put the file name in single quotes, so run.bat and debug.bat get run('cmat.m') again instead of run("cmat.m"). With the nested double quotes the tool got run(cmat.m). This is the same as the posix scripts and what it was before 991373c. Node labels already reject ', so the file name can't break out.

Added a test that builds a windows study with a .m node and checks run.bat and debug.bat. It fails on dev and passes with this change. I also ran the generated line with a script that prints its argv, and it now gets run('cmat.m'). Full suite passes.

CI on dev is red right now for two reasons unrelated to this, so I added them here to get the checks green:

• the C++ jobs fail because the comma tryparam test still writes in/1/concore.params after fix: read maxtime and params from in1 in concore.hpp #621. Same one-line fix as in fix: update s on ZMQ reads in Julia so unchanged() works #629.
• ubuntu-latest is now 26.04 and has no Python 3.9, so the test matrix fails to set up and cancels the rest. I pinned that job to ubuntu-24.04, since we still support >=3.9.

[Siddhanthguptaa]

Heads-up: #639 changes the same Windows run.bat/debug.bat lines in mkconcore.py, so the two PRs will conflict (merging both locally gives a conflict in mkconcore.py). They fix different things on those lines:

• this PR: run("x.m") → run('x.m')
• fix(mkconcore): launch quoted tool paths correctly in Windows run.bat and debug.bat #639: start /B /D "CZ" "<octave>" ... → start "" /B /D ... — since fix(mkconcore): handle tool executable paths with spaces in generated scripts #541 quotes the executable, start takes it as the window title, so Octave (and Python) nodes aren't launched at all.

So with only one of them, Octave still doesn't run from run.bat. With both applied (Windows 11, Octave 11.3, demo/sampleM.graphml), Octave starts and runs controllerM.m; on my machine it then stops at the first read with ym(1): out of bound 0 (dimensions are 0x0), which looks like a separate issue that I'm looking into.

Whichever lands second needs a small rebase. I'm happy to rebase #639 on top of this one.

[bhuvan-somisetty]

Thanks for testing both together. Yeah they fix different things on that line, so rebasing #639 on top of this works for me.

[Siddhanthguptaa]

Follow-up on the read error I mentioned above: the cause is that mkconcore doesn't copy concore_literal_eval.m / concore_literal_serialize.m into generated studies, so concore_read can't parse anything (#648, fix in #649). With those copied, demo/sampleM.graphml runs the full loop on Windows together with this PR and #639.

[bhuvan-somisetty]

Nice, good to know it runs end to end with all three. Thanks for tracking that down.