Commit 107eddda authored by Roger Dingledine's avatar Roger Dingledine
Browse files

declare that we're going to put 0.2.0.33 out this month.


svn:r17895
parent 363d9b21
Loading
Loading
Loading
Loading
+9 −9
Original line number Diff line number Diff line
Changes in version 0.2.0.33 - 2009-??-??
Changes in version 0.2.0.33 - 2009-01-??
  o Major bugfixes:
    - When a stream at an exit relay is in state "resolving" or
      "connecting" and it receives an "end" relay cell, the exit relay
+1 −20
Original line number Diff line number Diff line
@@ -3,36 +3,17 @@
description of the patch.)

Backport for 0.2.0:
  o r17135: ClientDNSRejectInternalAddresses not consistently obeyed.

Backport for 0.2.0 once better tested:
  o r16136: prevent circid collision.  [Also backport to 0.1.2.x??]
  o r16558: Avoid mis-routing CREATED cells.
  Xo r16621: Make some DNS code more robust (partial; see also libevent
            approach). (Also maybe r16674)
    [Partially backported.  Instead of the basic name checking, I backported
     r17171 instead, to be even more resistant to poisoning.]
  o r17091: distinguish "no routers support pending circuits" from
            "no circuits are pending."
    o See also r17181...
    o ... and r17184.
  o r17137: send END cell in response to connect to nonexistent hidserv port.
  o r17138: reject *:* servers should never do DNS lookups.
  o r17139: Fix another case of overriding .exit choices.
  o r17162 and r17164: fix another case of not checking cpath_layer.
  - r17208,r17209,r7211,r17212,r17214: Avoid gotterdammerung when an
    authority has an expired certificate.
  o r17562: Fix bug 874, wherein a sighup would make us kill all our intro
    points and leave their corpses for the next set of intro points to
    stumble over.
  o r17566: FIx bug 691, wherein failure to run dns_init() would kill Tor
    dead.
  - r17886:  Don't remove routerinfos as unlisted unless we have a
    consensus.
  - r17887:  Don't accept 1.2.3 as a valid IP address.


Backport for 0.2.0, maybe:
  o r14830: Disable TLS compression.  This saves RAM and CPU, and
    makes our TLS compression harder to distinguish from
    firefox's. Win/win/win, right?