git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Bloom filters for have/want negotiation

From
Junio C Hamano <gitster@pobox.com>
Date
Sep 11, 2015, 21:42 UTC
Message-ID
<xmqqfv2kzl91.fsf@gitster.mtv.corp.google.com>
In-Reply-To
<55F343F5.6010903@alum.mit.edu>
Michael Haggerty <mhagger@alum.mit.edu> writes:
Show 7 quoted lines
> 1. The server advertises the references that it has in the way that it
> is currently done.
> 2. The client advertises the objects that it has (or some subset of
> them; see below) via a Bloom filter.
> 3. The server sends the client the packfile that results from assuming
> that the Bloom filter is giving correct answers. (This might mean that
> too few objects are sent to the client.)

Wouldn't this end up sending objects the server has (perhaps reachable from a branch that is not advertised to the client) that is not reachable from the tips the client asked upon a false hit? If so, that has security ramifications for servers who restrict what you can download based on who you are (and which branches are supposed to be visible to you).

Using bloom filter (perhaps invertible ones) to identify what the receiving end lacks is certainly an intriguing approach, though.

Previous: Michael HaggertyNext: Shawn Pearce
Message 2 of 6 in “Bloom filters for have/want negotiation”
  1. Michael HaggertySep 11, 2015
  2. Junio C HamanoSep 11, 2015
  3. Shawn PearceSep 12, 2015
  4. Junio C HamanoSep 12, 2015
  5. Shawn PearceSep 12, 2015
  6. Michael HaggertySep 14, 2015

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.