Conversation
…ult download on UNIX)
|
Can one of the admins verify this patch? |
|
It seems most of the http_archive are really just github archive, can you point me to where did we solve the bintray issue? |
|
It's the mirror here :) f9b23a9#diff-5d1661b10448a2754e2c6c9ecd896d8bR34 |
|
any official documentation on the existence of mriror.bazel.build? who manages it? |
|
I don't think it's officially documented, but it's used by Bazel itself for lots of downloads, as well as repos like |
simon-mo
left a comment
There was a problem hiding this comment.
looks good, had few questions
.github/workflows/main.yml
Outdated
|
|
||
| env: | ||
| LLVM_VERSION_WINDOWS: 9.0.0 | ||
| GITHUB: true |
There was a problem hiding this comment.
maybe just use github's default env var? https://help.github.com/en/actions/configuring-and-managing-workflows/using-environment-variables#default-environment-variables
There was a problem hiding this comment.
Oh man I was looking for these, couldn't find them. Thanks!
| strip_prefix=True, url=None, path=None, **kwargs): | ||
| def urlsplit(url): | ||
| """ Splits a URL like "https://example.com/a/b?c=d&e#f" into a tuple: | ||
| ("https", ["example", "com"], ["a", "b"], ["c=d", "e"], "f") |
There was a problem hiding this comment.
can we use dict? https://docs.bazel.build/versions/2.0.0/skylark/lib/dict.html
There was a problem hiding this comment.
I can if you want, but I thought it's nicer to keep them in order and somewhat similar to Python's own urlsplit (which returns a tuple).
There was a problem hiding this comment.
yeah but python returns named tuple which are a bit more readable when retrieving the result.
There was a problem hiding this comment.
Haha yeah true, ok sure I'll change it.
| def auto_http_archive(*, name=None, url=None, urls=True, | ||
| build_file=None, build_file_content=None, | ||
| strip_prefix=True, **kwargs): | ||
| """ Intelligently choose mirrors based on the given URL for the download. |
There was a problem hiding this comment.
one sentence about the heuristic we are using here? (when to use bazel mirror and when to use github directly?)
There was a problem hiding this comment.
Currently the heuristic is we prioritize mirror.bazel.build URLs below GitHub URLs, but above everything else.
(The line you're looking for is prefer_url_over_mirrors = is_github.)
|
Test FAILed. |
|
Test FAILed. |
|
Test PASSed. |
This reverts commit 44aded5.
This reverts commit b6359fe.
* Add a test for LRU fallback * Update error message * Upgrade arrow to master * Integrate with arrow * Revert "Bazel mirrors (#7385)" This reverts commit 44aded5. * Don't LRU evict * Revert "Revert "Bazel mirrors (#7385)"" This reverts commit b6359fe. * Add lru_evict flag * fix internal config * Fix * upgrade arrow * debug * Set free period in config for lru_evict, override max retries to fix test * Fix test? * fix test * Revert "debug" This reverts commit 98f01c6. * fix exception str * Fix ref count test * Shorten travis test?
* Switch to mirrors.bazel.build where possible * Switch from .zip to .tar.gz for smaller downloads (it's also the default download on UNIX) * Use direct GitHub URLs in Bazel files for clarity * Don't pass patches to local_repository * Remove github_repository() * Switch to GitHub actions/checkout@v2 which is faster * Use faster extraction method for LLVm on Windows * Move LLVM_VERSION_WINDOWS to the shell script since it's not a CI-specific value * Change GITHUB_TOKEN to GITHUB * Don't show timestamps for GitHub Actions * Factor out some options from GitHub Actions * Tell Bazel to stay on the same volume in GitHun Actions * Display progress output when downloading toolchains Co-authored-by: GitHub Web Flow <noreply@github.com>
Why are these changes needed?
dl.bintray.comkeeps returning forbidden, so we want to be able to use a mirrorImprove build performance
Make build rules clearer and more succinct
Checks
scripts/format.shto lint the changes in this PR.