Skip to content
Notifications
Clear all

Anyone else find the file editing workflow clunky compared to inline edits?

1 Posts
1 Users
0 Reactions
5 Views
(@sre_night_owl_88)
Active Member
Joined: 4 months ago
Posts: 10
Topic starter   [#1406]

Just wrapped up another 3 AM war room—false alarm, of course—and I was using Aider to document the post-mortem. It got me thinking. The whole back-and-forth, file-by-file approval workflow feels like it was designed for someone who's never had to context-switch twelve times before their first coffee.

I'm talking about the `git diff` style edits. You get a block of changes, you have to say yes/no, maybe tweak the plan. During an incident, or even just a refactor, I'm already mentally juggling ten things. Now I'm playing spot-the-difference? I keep wishing I could just point at a line and say "change this variable name" or "add a log line here" like I can with Copilot directly in my editor.

```python
# Instead of this whole dance:
# Aider proposes:
- timeout = config.get('timeout_sec', 30)
+ timeout_seconds = config.get('timeout_sec', 30)

# I have to review & accept.
# What I *want* is to just write a comment:
# aider: rename timeout to timeout_seconds for clarity
# and have it happen inline.
```

It's powerful for large-scale rewrites, I'll give it that. But for the daily grind—quick fixes, tweaking configs, updating a deployment manifest—the friction is real. The chat feels like a separate layer, divorced from the code.

Anyone else feel this way? Found a workaround or a flag I'm missing? Or is this just the price we pay for the git-safe, audit-trail model? Maybe I'm just spoiled by inline edits after too many on-call shifts.



   
Quote