search

darshan-hpc / darshan

Darshan I/O characterization tool
Other
57 stars 28 forks source link

propagate version string update #882

Closed wkliao closed 1 year ago

wkliao commented 1 year ago

This simplifies the future version string update, by just editing file ./darshan.version.

tylerjereddy commented 1 year ago

To be clear, a Python developer should be able to iterate on the Python code, and depend on its static version string, without caring about any of the autotools/autoconf stuff.

wkliao commented 1 year ago

I can remove the 2nd commit which involves pydarshan.

From my experience on netcdf, I think making separate packages, i.e. darshan runtime, darshan util, and pydarshan is a mistake, because it allows users to mix different versions of the packages, i.e. use darshan-runtime version 3.4.0 together with darshan-util 3.3.0. There may be unexpected results.

shanedsnyder commented 1 year ago

Changes look good to me, this will simplify releases a bit.

I agree with limiting any sort of dependencies between PyDarshan and the traditional Darshan build system -- we've made this a point in the past as it seems to give us needed flexibility to accomplish what we want on both ends. Ultimately it would be nice if we could automate how PyDarshan captures these versions (as well as C headers in include), but I'm not sure we need to do it by adopting Autotools in PyDarshan -- it doesn't seem to be used generally in Python projects.

tylerjereddy commented 1 year ago

Ok, the Python stuff was removed, Shane approved, and you got a code review from an experienced build system developer (Eli) as well, so I merged it.

Incidentally, the CI only ran for Python on MacOS M1 native, I'm not actually sure what CI we'd want to flush for these changes normally.

shanedsnyder commented 1 year ago

Re: CI tests, I added the [skip actions] key word on my commit since it just modified a doc. I guess that annotation doesn't affect the Cirrus stuff?

tylerjereddy commented 1 year ago

Ah understood, yeah the services have different flags. Another project I work on summarizes some of them here, though I believe for some they added some custom handlers we could copy if we wanted https://scipy.github.io/devdocs/dev/contributor/continuous_integration.html#skipping