You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm not sure we want to depend on mkoctfile to install our package but I certainly don't mind if development depends on it...
At least Fedora (and probably others) runs tests as part of packaging. Our content is currently noarch but because of tests, we'd have compiled objects.
one option is to not have it as part of our formal tests.
In practice, probably only matters that CI runs it (as @catch22 highlighted).
@mtmiller: thoughts on this? Is it reasonable to compile the .oct code at run time, from within Octave?
The text was updated successfully, but these errors were encountered:
With my developer hat on, yes, if you have no problem with maintaining some extra C++ functions for CI testing, that sounds good to me. I suppose this could be a separate make target?
With my Debian packaging hat on, I think it would be nice to at least make it possible for distributions to run these tests if they want to. IOW it's all there in the release if they want it, with an optional flag or a separate make target name or something.
Is it reasonable to compile the .oct code at run time, from within Octave?
Not sure what you are distinguishing against here. If you mean is it ok that the .oct files aren't compiled until the tests are invoked, that sounds fine to me. They are test data, there's no need to structure them as if they were normal functions to be built and installed as part of the package.
See #179 (comment)
I'm not sure we want to depend on mkoctfile to install our package but I certainly don't mind if development depends on it...
noarch
but because of tests, we'd have compiled objects.@mtmiller: thoughts on this? Is it reasonable to compile the .oct code at run time, from within Octave?
The text was updated successfully, but these errors were encountered: