![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
TMI: RUST_PATH and package scripts
Jun. 25th, 2013 10:33 pmToday I finished two pull requests! The first one implements searching for packages using directories given by the RUST_PATH environment variable. The second one makes package scripts more useful by allowing them to call the default build and install logic after doing whatever custom steps are necessary.
Both of these pull requests are incomplete -- there are parts of the original bugs they address that aren't implemented. I didn't implement using RUST_PATH to determine the target directory for installs, and I also didn't teach the rustc driver itself about RUST_PATH. I opened up separate bugs for those (#7402 and #7398).
Likewise, for the package-script patch, I assumed in rustpkg that package scripts only can implement a custom hook for the install command, not for any other commands. I opened up #7401 for that.
I'm not really sure, from a bug-tracker-philosophy point of view, whether it's better to submit several pull requests over time for the same bug, or close a given bug after I've done a significant amount of work on it and open up newer, more specific bugs. Something just feels better about having older bugs closed.
Alex Crichton did some work on making sure rustc uses LLVM in a thread-safe way, which will allow us to actually re-enable the rustpkg tests (!) So I'm looking forward to that being finished too -- there's a pull request on that pending, and then I think all that's left is manually wiping the LLVM build directories on all the bots so that they rebuild LLVM.
Both of these pull requests are incomplete -- there are parts of the original bugs they address that aren't implemented. I didn't implement using RUST_PATH to determine the target directory for installs, and I also didn't teach the rustc driver itself about RUST_PATH. I opened up separate bugs for those (#7402 and #7398).
Likewise, for the package-script patch, I assumed in rustpkg that package scripts only can implement a custom hook for the install command, not for any other commands. I opened up #7401 for that.
I'm not really sure, from a bug-tracker-philosophy point of view, whether it's better to submit several pull requests over time for the same bug, or close a given bug after I've done a significant amount of work on it and open up newer, more specific bugs. Something just feels better about having older bugs closed.
Alex Crichton did some work on making sure rustc uses LLVM in a thread-safe way, which will allow us to actually re-enable the rustpkg tests (!) So I'm looking forward to that being finished too -- there's a pull request on that pending, and then I think all that's left is manually wiping the LLVM build directories on all the bots so that they rebuild LLVM.