Jonas Chevalier
Jonas Chevalier
If you have a Matrix client we can also talk about it in this room: https://matrix.to/#/#devshell:numtide.com
I banged on it until it works: https://github.com/mayl/gcovDemo/pull/1 Would you mind adding a C dependency like "libz" as well to exercise this out more? And related PR: https://github.com/numtide/devshell/pull/102
I think it's because gcc is being used in the stdenv too much. The wrapped gcc makes a bunch of assumptions that are not valid outside. That's also why gcov...
Do you think this PR is worth pursuing? It would need to have some tests.
How do you handle when there are two shells loading the same environment, is the second one detecting that all the processes have started? For now, I have worked with...
If you use `direnv`, then it also acts as an interface between your shell and whatever environment variables nix-shell is providing.
I think $SHELL is more reliable because sometimes the shell gets overridden in the Terminal emulator configuration. On Linux you can also look at `/proc/$PPID/exe`. But yeah, those are all...
I mean as a source of information. But devshell would have to be modified to take that into account. It also opens a bunch of questions like what to do...
Sub-shells are the only way to go if we want something more than env vars (ie direnv). The easiest for that is to start the sub-shell using the entry point...
I'm doing something similar but use a Procfile executed by [hivemind](https://github.com/DarthSim/hivemind) to start the services. @yajo the last commit changed postgres so make sure to replace `postgres` with `start-postgres`