Published 2025-08-26
tag(s): #overblown-minor-annoyances #yell-at-cloud #emacs #programming
I have this command
to create a
PR using the GH CLI directly from Emacs.[1]
For the reviewers, I have to type the GH usernames, so what I did is setup completion with my
teammates users to make sure they are correct. But, I can still type in any other username,
just in case I have to add someone that I don't regularly interact with.
Well, just now, I had to manually type in a name. And I had to stop myself a couple times from grabbing the mouse to copy and paste it from the browser window. I had the name right there, I verified it, but I still felt insecure about it.
This reminded me of when I removed company-mode
(basically, IntelliSense) and then eventually all automatic completion from
Emacs.[2]
And also when I was in college and had to solve some equation and was doing even 9x5 in the
calculator. Like, sure, I already had it there and turned on, but I know that 9x5 is
45, why "verify" it?
I don't know if it is a compulsion only I have, or if eventually anyone with access to tools stops trusting their own eyes or memory. But it bothered me enough just now that I am writing this post...
It is somewhat similar to going back in shell history to find ls -la , when just
typing the command is fast and easy enough. Yes, we can have some tool fill the details for
us, or recall the previous input, but is typing it that much work?
I don't have a conclusion. I recall some article I read once about how our memory might end up
working differently over the generations, now that we all have access to all the knowledge in
the world in our pockets.
Same with completion in code I guess. And that might get exacerbated with AI - automagically
getting not only a function name, but the code for a whole operation. Will some people stop
remembering how to open a file in Python, because the boilerplate is generated each time by
tools?
Does it matter?