feat: make repo pulling by agents done in parallel#12272
Conversation
|
I'm starting a first review of this pull request. You can view the conversation on Warp. I completed the review and no human review was requested for this pull request. Comment Powered by Oz |
There was a problem hiding this comment.
Overview
This PR parallelizes cloning of multiple agent environment and global-skill repositories by adding a batched clone path, then separates cloning from repo registration during environment preparation.
Concerns
- No blocking correctness or security concerns found in the annotated diff.
Verdict
Found: 0 critical, 0 important, 0 suggestions
Approve
Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).
Powered by Oz
| failed=1 | ||
| fi | ||
| done | ||
| exit "$failed" |
There was a problem hiding this comment.
If cloning any repo fails, do we get any output? Especially for failed clones I think we need to see the output, but ideally we can show the output of every clone that happened in a way that's not interleaved. I think we might need to write output to a log and print it when they're done?
|
Do you have a screenshot of what it looks like on success or when a repo clone fails? |
seemeroland
left a comment
There was a problem hiding this comment.
Approving to unblock assuming it's been tested. Also just making sure this script will work on any common shell and not using any zsh or bash specific syntax?
Yep, this is all POSIX |

Description
This PR makes repo pulling in agent mode done in parallel rather than done sequentially. Oz environments with many repos can take a long time currently to start up, and much of that time is due to pulling in multiple large repos one after another.
Linked Issue
https://linear.app/warpdotdev/issue/REMOTE-1867/parallelize-environment-repo-cloning-to-speed-up-startup
Testing
./script/runScreenshots / Videos
Before:

After:

Agent Mode