Issue1312 durack1 readme.md update

[durack1]

fix #1312 - Update readme.md for 2.2.0 release

[durack1]

@jbeezley I wonder if there is a disconnect between the source the buildbot is picking up - this PR was issued against master and yet the error indicates Version: uvcdat-2.2.0rc1-42-gaa1a3da:

Successfully updated your environment to use UVCDAT
(changes are valid for this session/terminal only)
Version: uvcdat-2.2.0rc1-42-gaa1a3da
Location: /home/buildbot/buildbot/uvcdat-garant-linux-release/build/install
Found Baselines:
- /home/buildbot/buildbot/uvcdat-garant-linux-release/build/uvcdat-testdata/baselines/vcs/test_canvas_background.png
Comparing 'test_backgroundcolor_white.png' to '/home/buildbot/buildbot/uvcdat-garant-linux-release/build/uvcdat-testdata/baselines/vcs/test_canvas_background.png'...diff=234313.725490, New best!
All baselines failed! Lowest error (234313.725490) exceeds threshold (10.000000).
Saving image diff at 'test_backgroundcolor_white_diff.png'.

It seems all these PRs are failing.. @doutriaux1 @aashish24

[aashish24]

@durack1 its not that. We will fix in the next release. what happens is that when we tag a commit, we need to bring that commit as "s ours" or else master will see that tag as well. I think buildbot uses the git describe which use the last tag information.

[jbeezley]

It isn't buildbot. It is the setup_runtime.sh. It uses git describe. The version message contains the hash of the commit being tested: uvcdat-2.2.0rc1-42-g​aa1a3​da

$ git log --oneline -n 1 aa1a3da
aa1a3da fix #1312 - Update readme.md for 2.2.0 release

[doutriaux1]

@aashish24 let's bring the tag back into master. Yet another reason I really don't like the "release" branch. But we can debate this ad vitam eternam...

[durack1]

@jbeezley @aashish24 @doutriaux1 it seems a bunch of the test images aren't being downloaded to compare: https://open.cdash.org/viewTest.php?onlyfailed&buildid=3822441

I wonder if the branch checkout in the unmerged #1321 (https://github.com/UV-CDAT/uvcdat/blob/issue_1312_tag_repos/CMake/cdat_modules_extra/checkout_testdata.cmake#L31) will solve this?

[aashish24]

@doutriaux1 the problem is not related to release branch. Whether or not we have a release branch this problem the information will exists because of how git describe uses the information. If we want something else, we need to either give up git describe and use something else. Or use tag master.

[aashish24]

It might. I am working on a branch that will automatically follow the branch name in uvcdat for uvcdat-testdata.

[durack1]

@aashish24 can you kick off another buildbot test once you've pulled #1321 back into master? It seems it's getting close now as it's passing on garant..

[durack1]

@doutriaux1 the PYOPENSSL error is causing grief here https://open.cdash.org/viewBuildError.php?buildid=3822471 - is that fix you were working on in master yet?

[jbeezley]

It looks like you may need to install libffi-dev on test-laptop.

[aashish24]

Thanks @jbeezley I just installed it.

[durack1]

@aashish24 wanna kick off test-laptop again then?

[jbeezley]

Done

[durack1]

@jbeezley thanks! Now I hope it passes..

[durack1]

@jbeezley ok great, it's just got a slight issue with the test image (problems with right boundary being written and not in test case: https://open.cdash.org/testDetails.php?test=338045449&build=3822602)

[doutriaux1]

@jbeezley I fixed this in my netcdf branch. we do build libffi so that's not the issue, one issue if the cffi_deps that as an issue with upper vs lower.

[durack1]

@doutriaux1 the only remaining issue with this current build is the test image - as you can see at https://open.cdash.org/testDetails.php?test=338045449&build=3822602 it's got a column of pixels being written to the image (far right side) which is not in the test image: https://open.cdash.org/displayImage.php?imgid=381312

[doutriaux1]

seems like the widgets color are different too

[doutriaux1]

@aashish24 just realizing you merged the wrong one.... You should have merged the one going to release not master... now the release will break as soon we update baselines...

[aashish24]

@doutriaux1 which branch you are referring to?