Missing git hash for release versions prints warnings

[NicoWeio]

I am not entirely sure if this is an error on my side, but when working with a release version of NuRadioMC, I always get a warning due to Git not finding a repository to read a commit hash from.

$ python3
Python 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import NuRadioReco.utilities.version
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
>>>

The relevant call stack in my specific case is

from NuRadioReco.modules.channelBandPassFilter import channelBandPassFilter
↓
import NuRadioReco.framework.event
↓
from NuRadioReco.utilities import io_utilities, version

As per #924, the herein-imported version.py tries to read the git hash upon loading.

But if I'm not mistaken, this will never work with a release version – the pip-installed package does not come with a .git directory.

Upon researching this, I also found that some code tries to warn the user if no hash is specified but will never actually do so, as all exceptions are caught before in version.py:

NuRadioMC/NuRadioReco/framework/event.py

Lines 691 to 696 in 5d309bf

	try:
	commit_hash = version.get_NuRadioMC_commit_hash()
	self.set_parameter(evp.hash_NuRadioMC, commit_hash)
	except:
	logger.warning("Event is serialized without commit hash!")
	self.set_parameter(evp.hash_NuRadioMC, None)

Here are my suggestions of what we could do:

• check if we are in a release version before trying to access the hash (speedup!)
• bake in the hash somewhere in the release pipeline
• silence the error by redirecting Git's stderr

And we should do the following:

• make the hash parsing more reliable by using the porcelain git rev-parse HEAD command instead of git status
• fix the defunct warning mentioned above, either by removing or rewriting it

I'll gladly create a PR, but we first need to decide on how we intend to handle the hashes.

[fschlueter]

Hey, thanks for the bug report! This indeed seems to be not the desired behaviour. If you could provide a fix in form of a PR that would be awesome and very much appreciated.
I agree with all your suggestions (if we checked that we are on a release, we do not even have to call the function and wont get the git error). Instead of the hash we could than simply store the version (that might not be so nice but much simpler).