.bash_profile executed by REPL shell commands

[staticfloat]

This seems like bad behavior to me:

shell> echo test
Filesystem                       Size  Used Avail Use% Mounted on
/dev/mapper/elementary--vg-root  885G  309G  532G  37% /

 17:43:00 up 2 days,  8:04,  6 users,  load average: 0.44, 0.26, 0.15
homes last updated: Wed Nov 22 14:21:26 PST 2017
test

My ~/.bash_profile is setup to output some stuff when I start up a shell. Even if this is considered questionable shell ettiquette, I'm not sure we should be sourcing ~/.bash_profile for every shell command executed from the REPL, should we?

[rfourquet]

I don't have a ~/.bash_profile, only a ~/.profile which doesn't not seem to be sourced, but my .bashrc is, which makes every shell command slow to be executed. It's not my domain, but sourcing such a file for each and every command seems wrong to me.

[KristofferC]

Dup of #17332

[staticfloat]

Maybe a dup.... that seems to be more about slowness than the underlying cause, but in any case, I am surprised that Julia sources this file whereas most of the other programming language REPLs that I have used (e.g. IPython) do not.

[KristofferC]

The cause of the slowness was due to sourcing a shell file on every command. Can close that one since this have a more accurate title.

[staticfloat]

So the difference here is that IPython does not pass -i to bash when it executes its shell commands, whereas Julia does. I'm going to open a pull request to remove passing -i to the shell and we can discuss the reasons behind why we might want to do that in the PR.