Bugfix/fix coverage

[gabyx]

Some fixes.

• [x] old git installation in kcov -> fix git command to support all version
• [x] MOCK_DOWNLOAD in tests to be able to mock all downloads for locally testing (no artefacts over the master branch). coverage.sh uses always the MOCK_DOWNLOAD, because its not so relevant there. For exec-tests.sh ist opt-in!
• [x] check kcov output, all should pass now except 116 which needs fixes.

[coveralls]

Pull Request Test Coverage Report for Build 524

• 25 of 33 (75.76%) changed or added relevant lines in 4 files are covered.
• 16 unchanged lines in 2 files lost coverage.
• Overall coverage increased (+3.05%) to 87.149%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
base-template.sh	5	7	71.43%
uninstall.sh	4	10	40.0%
<!--	Total:	25	33	75.76%	-->

Files with Coverage Reduction	New Missed Lines	%
uninstall.sh	1	83.94%
install.sh	15	87.55%
<!--	Total:	16		-->

Totals
Change from base Build 515:	3.05%
Covered Lines:	1953
Relevant Lines:	2241

---

💛 - Coveralls

[gabyx]

I try to correctly mock the download but test 094 still downloads the master I think. pretty weird. will work on it later

[gabyx]

I try to correctly mock the download but test 094 still downloads the master I think. pretty weird. will work on it later

[rycus86]

Thanks a lot for looking into this! Last time I tried to update kcov, I found that lots have changed in the next version, so I gave up on updating it at the time.

[gabyx]

Last time I tried to update kcov, I found that lots have changed in the next version, so I gave up on updating it at the time.

Tried it too, but lets do that in another PR maybe...

[gabyx]

Now, all coverage works except 116 and I still dont know why, strange... kcov is doing strange things... even the output is scrambled, :-|. Really cumbersome...

[gabyx]

Is there any consense about using git rev-parse --commo-git-dir versus --git-dir. Whould we use everywhere --commo-git-dir Git has to many behaviors :-), not a simple tool :-P

[rycus86]

Now, all coverage works except 116 and I still dont know why, strange... kcov is doing strange things... even the output is scrambled, :-|. Really cumbersome...

Yeah, that output gets scrambled unfortunately, never really worked out how to fix it properly. If you look at the generated coverage report in HTML, you can see which lines were executed vs which ones did you expect to be executed.

Is there any consense about using git rev-parse --commo-git-dir versus --git-dir. Whould we use everywhere --commo-git-dir Git has to many behaviors :-), not a simple tool :-P

Nope, I have no personal preference, except whichever existed first, so we don't limit compatibility with older Git versions if not necessary.

[gabyx]

Ok. I think this PR is good to go. 116 is still a mystery. Kcov is strange because sometimes not the whole output is shown and it stops suddenly (or just because we cannot see the whole output) and its pretty unobvious where it stopped. I will see if I can detect that in the report. A set -x lets everything go nuts :-)... Updating is an option? If I have time I will look into this.

[gabyx]

https://travis-ci.org/rycus86/githooks/builds/645188064

[gabyx]

Maybe we can leave test 116 out of the coverage, just to make it build and let the users know the build is fine. 116 should be fixed, will look into it soon hopefully with more details.

[rycus86]

Added some comments, but looks good in general. We can get this in and leave test 116 as is for now, just look at the comments please. Thanks!

[gabyx]

I did some review cleanup, and changed every relevant point to --git-common-dir as its the right thing to do. And I have seen you already did some tests for git worktrees :+1:

Could you please review the last changeset. If you are happy, maybe you can directly merge it, without having me to squash it :-)

[gabyx]

I will also try to replace all cd "path" && git ... to git -C "path" quickly.

[gabyx]

All tests work. Covarage still the same.

[rycus86]

Thanks a lot!