Bash shell integration keeps direnv from working correctly

[tristan957]

I find that when the Bash shell integration is enabled, direnv stops working. Has anyone else experienced this issue?

$ direnv version
2.32.3

$ bash --version
GNU bash, version 5.2.32(1)-release (x86_64-redhat-linux-gnu)
Copyright (C) 2022 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>

This is free software; you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.

[tristan957]

@jparise bisected the offending commit to d1eb8cc

[tristan957]

Follow up commit since I lost bash completion with the previous one 😆 tristan957/dotfiles@fefa87a

[jparise]

The Bash manual has a good explanation of the startup files and when they're run: https://www.gnu.org/software/bash/manual/bash.html#Bash-Startup-Files

We attempt to implement that as faithfully as we can in ghostty.bash.

Most relevant to your case, when we're launched as a login shell, we source /etc/profile and then the first of ~/.bash_profile, ~/.bash_login, or ~/.profile. That's why it's common for users to source their .bashrc from their .bash_profile to get consistent behavior across both login and non-login interactive shells.

I'm having trouble telling if there's a problem directly with direnv or if that was a side-effect of some/all of your .bashrc not being sourced in the way you expected?

I'm also not yet seeing what's causing your issue in 6af1850. In terms of which files are sourced, it should be equivalent to the previous version of that code. Because I'm not able to reproduce what you're seeing, could you try making a few local edits to narrow down what's going wrong? A few things to try:

• Revert the $HOME -> ~/ changes, just to rule those out
• Add some judicious echo statements to verify which branches are being used and what files are being tested and sourced

I'll also keep an eye out for other reports.

[tristan957]

yeah. I'll try and continue the investigation on my side tonight. Hopefully it is something that will show itself easily!

I wish I knew why my setup is so particular. I seem to be the only one that ever has bash shell integration issues.

[jparise]

I wish I knew why my setup is so particular. I seem to be the only one that ever has bash shell integration issues.

But it makes you a valuable tester! 😉

My bash setup is here, in case there's anything that might be helpful to you there: https://github.com/jparise/dotfiles/tree/master/bash

[jparise]

#5196 is a step toward reverting #5075. If it doesn't improve things, and we don't learn anything new, I'll move on to reverting the rest.

[jparise]

Now that we've had multiple reports and identified which change introduced the regression, I've promoted this to an issue: #5206

[jparise]

#5249 reverts the source of the problems introduced in #5075. I believe the root cause of most people's issues was the change to function-level scope, which was a flawed attempt to provide a more hermetic environment in which to source the startup files.