Closed daniel-mohr closed 3 years ago
@daniel-mohr FYI I have just pushed some commits I made yesterday and forgot to push upstream, one of which makes a change so that the linting job should now pass. I will review this tomorrow.
@sadielbartholomew I updated/merged your new master into my branch make_platform_independent. Unfortunately the tests do not run again and I also do not find a way to initiate a re-run. Any ideas? But the merge was simple.
Thanks @daniel-mohr.
Unfortunately the tests do not run again and I also do not find a way to initiate a re-run. Any ideas?
With the current settings for the workflows, all (I think) having set:
the easiest way, though not particularly elegant, to re-trigger the jobs is to open and close the PR to hit the reopened
category, which is what I have done for a while (feel free to try it and see). If you know a more elegant but equally simple way to re-trigger the jobs then please do let us know and/or update that part of the workflows to effect the necessary change in Actions behaviour.
@sadielbartholomew I commented your 'code comments', but after a new/better commit I cannot find them. But anyway, due to my better solution (based on my comments in #31) they are obsolete.
I will react on your other comments later (tomorrow or ...).
Merging #32 (111b958) into master (eca4441) will increase coverage by
0.39%
. The diff coverage is100.00%
.
@@ Coverage Diff @@
## master #32 +/- ##
==========================================
+ Coverage 80.71% 81.10% +0.39%
==========================================
Files 2 2
Lines 679 677 -2
==========================================
+ Hits 548 549 +1
+ Misses 131 128 -3
Flag | Coverage Δ | |
---|---|---|
unittests | 81.10% <100.00%> (+0.39%) |
:arrow_up: |
Flags with carried forward coverage won't be shown. Click here to find out more.
Impacted Files | Coverage Δ | |
---|---|---|
cfunits/units.py | 81.22% <100.00%> (+0.40%) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update eca4441...111b958. Read the comment docs.
Thanks @daniel-mohr, that sounds good. Take your time, there is unlikely to be any major or indeed minor changes going into cfunits
across the next week, as far as I am aware, so there is little chance for significant merge conflicts etc.
Regarding the code: https://github.com/NCAS-CMS/cfunits/blob/eca4441db8a124f4fac40b5011b359e40871c940/cfunits/units.py#L2225-L2230
I'm also not glad about it. But I do not want to decide this. Let's wait for @davidhassell .
Adapting setup.py regarding windows: Should we only add "Operating System :: Microsoft :: Windows" or should we do "Operating System :: OS Independent"? (I think it is as independent as udunits is available -- it's python!)
Adapting setup.py regarding windows: Should we only add "Operating System :: Microsoft :: Windows" or should we do "Operating System :: OS Independent"? (I think it is as independent as udunits is available -- it's python!)
Operating System :: OS Independent
please, it is more concise than listing out all of the possibilities (Linux, MacOS and now Windows)!
@daniel-mohr, one small thing to mention, FYI: I am quickly putting in a PR to do some automatic sorting of import statements across some libraries, including cfunits
(#33), so there is the potential for a small merge conflict from that, though I see you aren't applying any imports at this point, so hopefully there won't be any at all.
'setup.py' is now adapted.
I have no intention to add further code to this PR. So, from my point of view a merge of this PR could be done.
Thanks @daniel-mohr. The new commits look good to me, with the setup.py
being adapted correctly etc.
Since @davidhassell is due back from leave tomorrow, and it would be good to have a second opinion on the changes here because the code touched in the units
module is quite critical to cfunits
working properly, I will do a final review later today (since you have changed approach) but wait to merge until David has signed this off too, likely in the next few days or so.
Hi @daniel-mohr and @sadielbartholomew - lot's has gone since I've been away - great. I'll have a look at the dtype issues today.
OK, if I can help you let me know!
By the way, I enhanced list_numpy_dtypes to show the type mismatches on i386 Linux platform.
Version 3.3.4 now available in PyPi. Thanks @daniel-mohr and @sadielbartholomew for your work on this. Much appreciated.
At least most of #31 should be fixed with this pull request. For the given test cases the solution should be fine. Some problems addressed in #31 are not solved.