Defer errexit to as late as possible #70
Conversation
At the moment, transcrypt immediately exits with a failure when run outside of a git repository, regardless of what command is given to transcrypt. This causes unexpected behavior when running commands like "help" or "version". This commit moves errexit to as late as possible, while still keeping it as a safe guard for the time being. Ideally, errexit would not be used in transcrypt at all.
|
Fixes #68 |
|
@elasticdog Could you weigh in on this when possible, please? |
|
Hey @apeschel! Thank you for the PR, and my apologies on taking so long to respond. I think I've come up with a cleaner solution to issue #68 without having to shuffle around the errexit timing. There were a lot of changes to enable "strict" Bash within #53 that took me a while to warm up to, but now I can see the value (especially since transcrypt doesn't have a test suite we can rely on, it can help when adding new features). Let me know if my other PR addresses the issues you were tackling here, and I'd also be curious to hear your thoughts on why you think errexit should be eliminated completely. Cheers. |
|
I'm going to close this one out, as the original issue was addressed via #71, but thanks again for taking the time to look at the problem...I'd be glad to discuss errexit with you further if you have strong feelings about it. |
At the moment, transcrypt immediately exits with a failure when run
outside of a git repository, regardless of what command is given to
transcrypt. This causes unexpected behavior when running commands like
"help" or "version".
This commit moves errexit to as late as possible, while still keeping it
as a safe guard for the time being. Ideally, errexit would not be used
in transcrypt at all.