The sqlite download currently downloads the bz2 archive to a temporary file, and then extracts the database from that local file. This is not ideal for two reasons:
it increases the disk needed, as there has to at least momentarily be enough disk space to hold the compressed archive and the uncompressed database
it increases the time needed to download. Ideally the file would be uncompressed as it is downloaded.
It appears this was done since the timestamp of the archived file is generated after it's downloaded, and used for future comparison to avoid re-downloading identical files in the future.
The sqlite download currently downloads the
bz2
archive to a temporary file, and then extracts the database from that local file. This is not ideal for two reasons:It appears this was done since the timestamp of the archived file is generated after it's downloaded, and used for future comparison to avoid re-downloading identical files in the future.
We could probably streamline this by using
curl
to get the remote last modified time via HEAD request, and then downloading the archive, without a temporary file, immediately after.