Introduction to history deltas

Aaron Bentley aaron.bentley at utoronto.ca
Wed Dec 7 14:26:54 GMT 2005


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Johan Rydberg wrote:
| Aaron Bentley <aaron.bentley at utoronto.ca> writes:

|>It's not a clear win.  With history deltas, your latency scales
|>per-revision; with file-knits, your latency scales per-file.

| Note that the idea with knits are that you do not fetch the whole
| file, but instead fetches the deltas from the knit file (using http +
| ranges).
|
| Which means that the number of fetches are O(deltas) and not
| O(modified files).  It can of course be optimized, but that is the
| basic idea.

That would be silly.  Unless you can batch the reads on a per-file basis
to avoid latency, it's probably cheaper to read the whole thing so you
have all the deltas you need for that file.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDlvEu0F+nu1YWqI0RAgheAJ96Bc5/+3rtS1UGrIvJJZGpucPmQQCffGEF
srxolimt8x36htPPibkIpos=
=QNp7
-----END PGP SIGNATURE-----




More information about the bazaar mailing list