Skip to content

Releases: GEOS-ESM/mepo

Clean up of mepo save

19 Mar 17:44
a393404
Compare
Choose a tag to compare

DOI

This release adds nicety to mepo save where it will add blank lines between the sections in the components-new.yaml file.

Fix for cloning

16 Mar 19:40
5fa4144
Compare
Choose a tag to compare

DOI

This is a bugfix for mepo. Before, mepo did this when cloning a branch/tag called foo:

git clone repo.git
git checkout foo

This worked just fine until an update to the MOM6 repo in GEOS. The issue is that in MOM6, the fork was updated and in that update, the submodules underneath changed on the default branch and there were less. So, in doing so, you couldn't get the new submodules. The solution is to do:

git clone -b foo repo.git

This exposed some other issues in status and compare which seem to be fixed, but time will tell.

Bug fix for mepo status

05 Jan 18:52
8581a48
Compare
Choose a tag to compare

DOI

This release fixes a bug in mepo status.

The way mepo was figuring out what branch you were on with detached head was:

git show -s --pretty=%D HEAD

and then taking the second field. In most cases, this is fine, but sometimes you have two branches that are identical:

git show -s --pretty=%D HEAD
HEAD, origin/feature/wputman/dyamond_v2, origin/develop, origin/HEAD, develop

So, even though you checked out develop, mepo status was returning that after running mepo clone you were on feature/wputman/dyamond_v2. While this isn't incorrect, it doesn't match where we started and mepo shows a difference.

After looking at the git source code, I think it uses the internals used for the reflog to get this. If I look at the git-reflog output:

git reflog
eb7e199 (HEAD, origin/feature/wputman/dyamond_v2, origin/develop, origin/HEAD, develop) HEAD@{0}: checkout: moving from develop to origin/develop
eb7e199 (HEAD, origin/feature/wputman/dyamond_v2, origin/develop, origin/HEAD, develop) HEAD@{1}: clone: from github.com:GEOS-ESM/GEOSgcm_GridComp.git

Now we see internally it knows we checked out develop and then detached to origin/develop. We can capture this knowledge with git reflog HEAD -n 1 and then take the last part of the string.

I also clean up an odd save issue.

Add mepo commit -a

20 Nov 16:56
78eaaee
Compare
Choose a tag to compare

DOI

This release adds the ability to do mepo commit -a or mepo commit --all a la git commit -a/--all. It essentially runs mepo stage and then mepo commit.

Thanks to @WilliamJamieson for this feature.

Fix no-config mepo clone

20 Nov 15:49
13651a2
Compare
Choose a tag to compare

DOI

The last release introduced a bug when doing mepo clone URL without --config. This fixes that.

Fix for using config with "all-in-one" clone

20 Nov 15:15
979fde2
Compare
Choose a tag to compare

DOI

This fixes a bug in doing:

mepo clone --config <config-file> URL 

Add --force flag for fetch and fetch-all

20 Nov 15:01
2cfbbce
Compare
Choose a tag to compare

DOI

Adds a --force flag to mepo fetch and mepo fetch-all

Add --staged option to mepo diff

12 Nov 14:34
2ef0081
Compare
Choose a tag to compare

DOI

This release adds a --staged option to mepo diff a la git diff --staged

Bug fix for Fixture Initialization

09 Nov 20:10
0f0859b
Compare
Choose a tag to compare

DOI

This fixes a bug in fixture initialization. mepo assumed all branches were detached head but fixture branches will not be (since we don't detach those)

Bug fix for CI Systems

30 Oct 15:25
7817d16
Compare
Choose a tag to compare

DOI

Some CI systems clone repos in odd ways (for GitHub Actions, it's not even a clone!). So, some fixes are needed for the weird edge cases of repository "getting".