maxfun=8000 for cobyla in `t_large.c`

[zaikunzhang]

Hi @jschueller ,

Is there a particular reason for setting maxfun=8000 for cobyla in t_large.c, but not others? Will it be OK if we simply use the same maxfun as other solvers in t_large.c?

BTW, would you mind if this file is renamed to stress.c?

Thanks.

Zaikun

[jschueller]

yes, I set it to 8000 so it doesnt exceed the CI 6h time limit (should run for ~5h according to my experimentations)

[zaikunzhang]

With https://github.com/equipez/github_actions_scripts/blob/de1b45b9fd667a9fe42cd7d5ecec4318f60d9117/check_success_timeout, it does not matter if the job takes more than 6 hours. It will be canceled and marked as a success.

[zaikunzhang]

I made the above-mentioned changes in https://github.com/libprima/prima/commit/eb5707c152b52f1b8061d89968d27786e994d831. I hope this is fine with you. We can still make changes if you would like.

BTW, I modified the value of maxfun in the C examples, in order to show to the user that it is sensible to set maxfun to a multiple of n.

In addition, I removed the initialization of nf in the C examples, as it will be set by the solver anyway. I hope this is not considered bad practice in C.

Tell me if any of the above points is problematic to you. Otherwise, you may close this issue.

Thank you.

[jschueller]

its fine

[zaikunzhang]

In addition, I removed the initialization of nf in the C examples, as it will be set by the solver anyway. I hope this is not considered bad practice in C.

Indeed, I was wrong here. It was not initialization but declaration. Corrected.