Closed prisae closed 1 year ago
Merging #93 (49065b5) into main (999d9ca) will decrease coverage by
0.56%
. The diff coverage is77.21%
.
@@ Coverage Diff @@
## main #93 +/- ##
==========================================
- Coverage 87.15% 86.59% -0.57%
==========================================
Files 4 4
Lines 366 388 +22
==========================================
+ Hits 319 336 +17
- Misses 47 52 +5
I'm in favor of this and not having release branches
Let's bring this in and see how it goes.
This PR is mainly meant for discussion. These would be the changes required to change to
setuptools_scm
.The advantage of this approach is that you never have to hard-code a version any longer. The version numbers are created when installing it (e.g.,
pip install -e .
or similar) and placed inscooby/version.py
.Example of version numbers:
If scooby is not properly installed, e.g., the folder is copied somewhere without the
.git/
-information, then the version will be unknown plus date:unknown-20220727
.If you don't like this approach we can close this PR again.
Either way, I just found out that this setup does not seem to work with the release-branch setup that was implemented in https://github.com/banesullivan/scooby/pull/90 - because installing it now results in a version number
v0.5.13.dev...
, so it does not recognize the tagReleases/v0.6.0
. This means, if we want to use this we would either have to revert the release branches approach, or adjust my implementation here. As in this approach there are never two commits with identical version numbers, I think we could simply get rid of the release branches again. Edit: To support release branches, we have to addto the
use_scm_version
-dict insetup.py
. But I guess first we decide if we keep release branches or not.There is one oddity that you might see when looking through the changes:
MANIFEST.in
is normally used to include files.setuptools_scm
reverts that, so everything is included by default, andMANIFEST.in
is used to exclude files.TODO
setuptools_scm
torequirements: host:
.