I've finnaly managed to run the test grid and compare the results \o/ :-)
I've investigated the differences (potential regressions) by hand, and it seems that they behave the same on both ECL versions (short summary):
| | 16.0.0-c | 16.1.2-c | | cl-openid | operation not permitted (OS) | operation not permitted (OS) | | clack-errors | connection refused | connection refused | | quickutil | all fine | all fine | |--------------+------------------------------+------------------------------| | | 16.0.0-b | 16.1.2-b | | drakma | segment-violation | segment-violation | | i-type | frame-stack-overflow | frame-stack-overflow |
Some differences are present regarding handling the conditions. 16.1.2 catches overflow better with a proper restarts instead of jumping straight into the toplevel, but the causing error is the same. Reports attached.
I still had some problems with the batch run of cl-test-grid and had some minor confusions (SBCL as a preffered lisp signalled a condition once during the run, something with FIND-SYMBOL – sorry, I didn't preserve the log unfortunately. I have SBCL 1.3.1 as a preferred lisp in #'make-agent.
Also, after running for four implementations only three results were submitted (the last one was cached though, so I hadn't to re-run the tests, it submitted it right away).
Best regards, Daniel
Anton Vodonosov writes:
16.02.2016, 23:09, "Daniel Kochmański" daniel@turtleware.eu:
OK, thanks. I think I have something wrong with my setup because I can't load the crashed packages even from SBCL or CCL (problems with hu.dwim.*).
How do you test that, via cl-test-grid or in REPL?
I'll create a fresh user account, download testgrid and run it (with a new rc tag), I hope it will work OK...
OK. BTW, when configuring RC results, you could specify "ecl" storage for them (instead of the default "main").
Like here: https://github.com/cl-test-grid/cl-test-grid/blob/master/run-agent.sample.li...
Let me know in case of problems.
Best regards,
- Anton