Gitleaks exits with exit code 0 when ambiguous argument to log-opts are provided

[JoostVoskuil]

Describe the bug
Gitleaks exits with exit code 0 when ambiguous argument to log-opts are provided

To Reproduce
Run Gitleaks with log-opts that triggers an error from git log

Expected behavior
When error occurs I would expect exit code 1

** Logging**
8:47AM ERR fatal: ambiguous argument '18ee779d989c28192f5d439a91bcbbd3bf11dac2^..ffc5f83a4f995ed146c27fa32227ddfff251f282': unknown revision or path not in the working tree.
8:47AM ERR Use '--' to separate paths from revisions, like this:
8:47AM ERR 'git [...] -- [...]'
8:47AM INF no leaks found
8:47AM INF scan completed in 30.9416ms

Basic Info (please complete the following information):

• OS:Osx Monterey
• Gitleaks Version: 8.0.4

[zricethezav]

I'm not able to reproduce this behavior:

➜  ~/code/gitleaks (master) ./gitleaks detect --source=. --log-opts="18ee779d989c28192f5d439a91bcbbd3bf11dac2^..ffc5f83a4f995ed146c27fa32227ddfff251f282"

    ○
    │╲
    │ ○
    ○ ░
    ░    gitleaks

8:23AM ERR fatal: ambiguous argument '18ee779d989c28192f5d439a91bcbbd3bf11dac2^..ffc5f83a4f995ed146c27fa32227ddfff251f282': unknown revision or path not in the working tree.
8:23AM ERR Use '--' to separate paths from revisions, like this:
8:23AM ERR 'git <command> [<revision>...] -- [<file>...]'
➜  ~/code/gitleaks (master) echo $?
1
➜  ~/code/gitleaks (master)

This chunk of code is responsible for catching errors coming from git https://github.com/zricethezav/gitleaks/blob/master/git/git.go#L73-L85

I suppose what could be happening is the goroutine that continues the main execution finishes before the error can be scanned.

[zricethezav]

Doing a little more testing and I'm realizing there is no reason to open a goroutine for streaming errors. The error checking should happen once at the beginning and that's it.

Just kidding, stdout is blocked if I remove the goroutine invocation. A simple hack would be to add a sleep timer for like 100 ms which would give the stderr scanner enough time to check for errors. I think I'll introduce that

[zricethezav]

@JoostVoskuil please try v8.0.6 and see if this is still an issue

[JoostVoskuil]

Awesome, can't reproduce this anymore. You are really fast @zricethezav , can't keep up haha

[savely-krasovsky]

@zricethezav maybe we should just use channel to sync instead of using time.Sleep hack?

[zricethezav]

@L11R that would be ideal! Once upon a time I tried to implement that but failed. Happy to accept a community PR if someone wants to give it a shot

[savely-krasovsky]

@zricethezav I've prepared a draft PR: #1250! Also would be nice if you review those ones: #1242 and #1249