Add fsspec support to index remote filesystems, separate package / cache, database / output directories into 3 places. · Issue #143 · conda/conda-index · GitHub
More Web Proxy on the site http://driver.im/
You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Allow an optional fsspec dependency to index remote packages through the API. When fsspec is not installed local packages can still be indexed as usual.
Separate input, cache (sqlite databases), and output into three directories. By default these can be the same folder, but with fsspec the input will typically be read-only and not usable for the sqlite cache database.
Why is this needed?
This will be used to improve the CDN process.
The text was updated successfully, but these errors were encountered:
These changes look like they will help me; I'm currently working on creating an index in s3 and CloudFront. Are there plans to add channel_url to the CLI?
That's great! It would be interesting to add channel_url to the CLI. One challenge is that it would not be completely general; for example if the remote filesystem is HTTP then conda_index might call HEAD on each individual remote file to get the equivalent of mtime; those details might need to be stored out-of-band by overriding save_fs_state or by relaxing the comparison in changed_packages to only consider metadata that we could efficiently gather from upstream. On the other hand a s3 list would be reasonably efficient for small or medium sized channels.
Uh oh!
There was an error while loading. Please reload this page.
What is the idea?
Allow an optional fsspec dependency to index remote packages through the API. When fsspec is not installed local packages can still be indexed as usual.
Separate input, cache (sqlite databases), and output into three directories. By default these can be the same folder, but with fsspec the input will typically be read-only and not usable for the sqlite cache database.
Why is this needed?
This will be used to improve the CDN process.
The text was updated successfully, but these errors were encountered: