Conversation
|
Thanks, man! That was fast. I am a bit worried that I hope that makes sense. |
|
A gentle ping. |
|
Hrm. Sorry I missed your comment. I'm reading through now and trying to work out what's the best solution. It's pretty hacky but I'm tempted to just attempt to make final_session_setup idempotent. I can mostly see what the refactor the option parsing needs would look like but unfortunately I don't have time. |
|
Let me know what you think of these patches |
|
Thanks! I think the solution is close, but it won't work in some cases. We invoke |
|
So this kinda seems like an idiosyncrasy of ruby, in that it seems like invoking pry more than once in a single VM is probably going to be pretty dicey. should it reload all your plugins every time? Given that there's not any obvious way for plugins to deinitialize themselves it seems unlikely. That said, I can add a shim to the tests that will just munge that variable back to |
|
I forgot to mention that we actually use Anyway, what you said is a very good point. It's indeed tricky, that's why we use the require 'pry'
Pry.start(self, {})
In my opinion the most optimal solution is to call |
|
Ok, I think this should be good to go! Thanks for your help sheparding! |
|
Contributor of the week! 🎉 |
cc @kyrylo
I tracked down the bug, it's that loading rc files and actually processing the impact of (rc files + flags) were linked.
This fixes that. I kinda poked around trying to work out how to get some tests for this, but came up blank. Let me know if it looks obvious. I did some regression testing in a shell and in an internal project that uses pry, and it looked ok.