opencode: 0.0.52 -> 0.1.194#419604
opencode: 0.0.52 -> 0.1.194#419604GaetanLepage merged 8 commits intoNixOS:masterfrom delafthi:push-royomvsvxpkl
Conversation
|
mind updating to v0.1.136? |
GaetanLepage
left a comment
There was a problem hiding this comment.
Thanks for handling this non-trivial update!
|
It's unfortunate we have to fetch the prebuilt binaries... |
|
so i think what's adding to the difficulty is we have our cloud stuff in the same repo. opencode itself does not use sharp it's the remote stuff that does i'm wondering of |
|
I currently don’t have access to the logs. @thdxr, thanks for pointing that out. I'll try this and provide feedback as soon as I have an update and also update to the most recent opencode version. |
|
I am trying to use this derivation, but the binary is just bun. Each time I run Doesn't the opencode project provide the actual project in its releases? Why would they package bun?! |
I’ll try to push an update tomorrow to build opencode from source instead of just installing the binaries, so probably this issue should be resolved then. |
|
@thdxr A few notes about the package declaration, which is a bit more complicated:
|
|
@GaetanLepage I also resolved your previous comments about the short description and the version check. |
I'm getting this, not sure why? Does it work on your side? |
|
Yes, the hash of It might be related to the system architecture. For reference, I built on both |
|
this one was on x64 Linux |
|
Indeed, the hash value differs across systems. I'm not sure why it worked on my aarch64-linux setup with the same hash. The latest update should resolve this issue. |
|
The output hash of |
|
don't know how this gets created in upstream, but can't we get those node modules some other way? changing 6 hashes on package upgrade at this point looks wonky haha |
I agree. Let's hope for |
|
|
Added the following changes:
|
- Replaced `stdenv` with `stdenvNoCC` - Moved subpackages into main derivation to enable updating subpackages with `nix-update-script` - Fixed version test, by providing a temporary `HOME` directory
|
Fast-moving target... already at 0.1.183... |
|
|
is there a nixpkgs policy about update frequency? opencode releases a lot of versions, mostly small changes. should be update every x days at a maximum or something? |
|
not that I've seen, but I could be wrong. it's just maintainer (and reviewer) burden 😄 |
|
at some point, you have to commit! |
sure, I mean after this PR. now merging x or x+1 is not really important |
|
well... I think everything inbetween ~0.1.172 and 0.1.183 would take like 5 minutes before being able to properly post a message when using github copilot auth. so it depends on maintainer's urgency I guess. but people can always overlay/override locally unless dependencies or build change. |
|
On the current version, I'm getting the following build error on both x86_64-linux and aarch64-darwin:
I observed that in v0.1.185:
I did a |
|
Yes, I have the same issue and it seems that there is no way to ignore that error and just build based on the lockfile. |
Upstream updated the lockfile. Thus, this issue is fixed in 0.1.194. |
|
|
Summary
opencodefrom version0.0.52to0.1.126. This update includes several important changes:Details
sharpdependency to compile, and I couldn’t find any existing references or solutions to this problem. As a result, using prebuilt binaries is the current workaround until a better build process for Bun-based packages is available.Notes
I'll also add a home-manager module -> see nix-community/home-manager#7320
Please let me know if you need any additional information or testing!
Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.