Submodules not cleaned

[Niklas Angare <li51ckf02@…>]

regression.py cleans the source directory of the super-project, but not the submodules. The "git status" command it runs just after cleaning shows this for me:

On branch develop
Your branch is up-to-date with 'origin/develop'.

Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)
  (commit or discard the untracked or modified content in submodules)

        modified:   libs/interprocess (modified content)
        modified:   libs/numeric/odeint (untracked content)
        modified:   tools/build (modified content, untracked content)

no changes added to commit (use "git add" and/or "git commit -a")

The changes in libs/interprocess were caused by my patch_boost script. The changes in libs/numeric/odeint were generated files. The changes in tools/build were generated files plus manual changes I made to test something between runs.

Adding the following command would appear to fix it:

git submodule foreach "(git reset --hard; git clean -fxd)"

While researching this I stumbled upon the fact that orphaned submodule directories aren't removed by git clean unless you pass it -ffxd (see the help for the -d option). However this will remove submodules that are present in another branch but not the current one, which may or may not be a good idea.

Regards

Niklas Angare

[Niklas Angare <li51ckf02@…>]

After discussion on the testing mailing list ("Submodules not cleaned"), this was fixed on the develop branch (commit ac7fc73f).

It later turned out that the quoting used in the Python code gave an error with Git 1.9 on Windows (testing list "New Windows Runner reset hard error"). I'm attaching a patch that fixes it. It works for me with both Git 1.8.3 and 1.9.2 on Windows and Git 1.9.2 on QNX and for Tom Kent on Windows.

[Niklas Angare <li51ckf02@…>]

Fix for unknown option hard

[Thomas Kent <teeks99@…>]

I've got a runner that can't be set to run automatically until this patch is committed.

[anonymous]

Just pushed this to develop.

[Thomas Kent <teeks99@…>]

I see that it was pushed to develop, then backed out again? Was there a problem?

[Noel Belcourt]

I was afraid I'd broken some testers so I reverted it. It now seems that was unnecessary so I'll re-apply the patch.

[René Rivera]

Replying to noel_belcourt:

I was afraid I'd broken some testers so I reverted it. It now seems that was unnecessary so I'll re-apply the patch.

Did you re-apply the patch? Just wondering if we can close this.

[René Rivera]

Never mind.. I see that it's been applied and it's in the new regression repo version. So closing.