[tor-bugs] #25667 [Core Tor/Tor]: LZMA/ZSTD descriptor compression support

Tor Bug Tracker & Wiki blackhole at torproject.org
Thu Mar 29 17:30:59 UTC 2018


#25667: LZMA/ZSTD descriptor compression support
-------------------------------------------------+-------------------------
 Reporter:  atagar                               |          Owner:  (none)
     Type:  defect                               |         Status:  new
 Priority:  Medium                               |      Milestone:  Tor:
                                                 |  0.3.3.x-final
Component:  Core Tor/Tor                         |        Version:
 Severity:  Normal                               |     Resolution:
 Keywords:  regression? 033-must needs-analysis  |  Actual Points:
Parent ID:                                       |         Points:
 Reviewer:                                       |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by atagar):

 Thanks Nick!

 > If not, this might not be 033-must, but it would be a shame to ship with
 a bug like this.

 Agreed. To be clear I don't think this is particularly important. The
 introduction of lzma and zstd support are nice additions but if they don't
 work then not the end of the world. :)

 > First off -- just because Moria is 0.3.4, doesn't mean it will
 necessarily support zstd and/or lzma2. It only supports those if it
 detects them at compile time

 Ahhh. My first thought when I read this was "how is this a useful
 feature?", but in reading proposal 278 I see now that the caller is
 supposed to provide *all* compression schemes they're willing to accept.

 I think we need to change the spec in a few ways...

 1. The spec should say that relays 'MUST' support plaintext and
 deflate/gzip (ie. callers can rely on those) but that relays only 'SHOULD'
 support lzma and zstd.

 2. Callers should advertise all compression schemes they support in
 'Accept-Encoding'. The relay will then pick the best it supports from
 among those (falling back to identity if none are supported). The
 compression scheme used is the response is indicated in the reply's
 'Content-Encoding' header.

 3. The spec should say how we pick among the compression. That is to say,
 lzma > zstd > gzip > identity (or whatever the actual behavior is).

 If this sounds good and you can tell me how tor actually picks for #3 I'll
 send ya a spec patch.

 Cheers! -Damian

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/25667#comment:2>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list