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
We did a bunch of improvements to make linking files faster. There are some speedups to be had when downloading files as well:
I think we should download into a "spooled temporary file". Currently we are always downloading and immediately extracting the file. Extraction can be slow, which slows down the download as well. IMO we should do both as fast as possible, extracting and downloading.
We also use the default BufReader of bzip2 / zstd when decompressing. While this is perfectly fine for files loaded from disk, it might not be optimal for downloads. The default buffer size is 8 KB. E.g. uv is using 128KB buffer size instead. So I think increasing the buffer size might result in faster downloads & faster decompression as well.
The text was updated successfully, but these errors were encountered:
We did a bunch of improvements to make linking files faster. There are some speedups to be had when downloading files as well:
I think we should download into a "spooled temporary file". Currently we are always downloading and immediately extracting the file. Extraction can be slow, which slows down the download as well. IMO we should do both as fast as possible, extracting and downloading.
We also use the default BufReader of bzip2 / zstd when decompressing. While this is perfectly fine for files loaded from disk, it might not be optimal for downloads. The default buffer size is 8 KB. E.g.
uv
is using 128KB buffer size instead. So I think increasing the buffer size might result in faster downloads & faster decompression as well.The text was updated successfully, but these errors were encountered: