Why not source-ing the $NVM_PROFILE after install?

[IonicaBizau]

Instead of reopening the terminal I did source ~/.bashrc -- which could be added to the install script (install.sh):

source $NVM_PROFILE

If you like the idea, I can create a pull request.

[ljharb]

It's not a bad idea.

Although, if we simply want nvm to be active immediately, we could just directly source nvm.sh. Perhaps that'd be better than forcing a re-source of $NVM_PROFILE, which may not always be set?

[IonicaBizau]

I agree. Even better! This would look like source $NVM_DIR/nvm.sh, right? 🍀

[ljharb]

yes. it'd be . $NVM_DIR/nvm.sh to be more portable, altho the install script is always bash so it's OK either way.

[IonicaBizau]

@ljharb Thanks! I opened #776. 😄

[ljharb]

@IonicaBizau Turns out this couldn't possibly work, because install.sh is executed, and piped to bash - not sourced directly.

[IonicaBizau]

@ljharb Hmm, from what I remember it worked fine for me when I installed it on a different machine. Maybe can it be improved/fixed?

[ljharb]

How did you install it on that machine? The install command in the README curls install.sh and pipes it to bash - which won't be the current shell.

[IonicaBizau]

@ljharb Confirming this doesn't work. Just tested it. I have in mind two possible solutions:

• downloading the script and executing it as file (probably then the source will work)
• adding the source call in the installation command (e.g. curl ... | bash && source ???)

...sorry for this bug. 😢 I'm happy to fix it.

[ljharb]

No worries - it's been reverted.

Adding a . $NVM_DIR/nvm.sh line to the command also wouldn't necessarily work, since NVM_DIR won't yet be available in the current shell.