On buildbot we currently rely on https://cs.chromium.org/chromium/build/slave/run_slave.py?q=run_slave.py&sq=package:chromium&dr=C&l=168 to clean up unwanted directories in between builds. To this end, we should:
  * Add <build_root>\b\tmp\* to the list of paths to delete in this function
  * Add C:\tmp\* to the list of paths to delete in this function
  * Change {remote,annotated}_run to set 
    * b\tmp\<new_random_dir> as temp_build
    * C:\tmp\<new_random_dir> as temp_system
    * (on linux/mac we'd use <build_root>b/tmp/<random> for both)

Recommend keeping "tmp_base" as-is as the system-side temporary directory ("tmp_system"  here).

Let's replace "cleanup" with "tmp_build", which is a build-filesystem temporary directory.

why not rename tmp_system? It would make it more obvious what its purpose is on initial inspection.

er... `why not rename tmp_base to tmp_system`.

Thinking just b/c it's referenced a lot.

I'm not worried about changing the references. I've become pretty good at that :)

From reading the bug description it is still unclear why system temp is needed. Why?

Answering for iannucci@, but apparently some Windows programs have the expectation that the tempdir used and exported be on the same filesystem as the user profile.

Yes, there are some tests (like the installer tests) that assume they can do same-volume rename operations, and they must install on the system drive (i.e. C:). This is less of a problem but still applicable on linux.

"System" is somewhat reductionist, but I think it should be enough to cover our uses for now.

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot