Timeline
Feb 9, 2010:
- 9:49 PM Changeset [f61e7ac]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Don’t allocate 32 times as much memory as we need for decrypting. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [1f37eba]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Don’t read off the end of misaligned input messages. This fixes an input buffer overflow noticed by asedeno and andersk. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [9be80b5]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Add tests. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [ae50abc]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Don’t append a newline when decrypting. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [a52eeb1]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Add separate functions for zcrypting with a particular key. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [dacb555]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Change zcrypt functions to allocate their output buffer. The logic to determine the right buffer size is complicated and does not belong in callers. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [d35d2ba]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Don’t leak the body after decrypting a message. This fixes a leak noticed by nelhage. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 9:49 PM Changeset [3f18ad5]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Fix zcrypt when compiling without krb4 OpenSSL's "compatibility" DES API makes des_key_schedule a structure containing an array, while in krb4 it was an array directly. Thus, we need to pass a key schedule pointer to owl_zcrypt_string_to_schedule, instead of passing the key schedule by value. The compatibility macros defined in OpenSSL's des_old.h allow that function's passing of *schedule to des_key_sched() to DTRT. Reviewed-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
Feb 6, 2010:
- 6:11 PM Changeset [9a023d0]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Send multiple PRIVMSGs for IRC messages entered as multiple paragraphs It is occasionally useful to be able to send multiple lines of IRC text at once, and I've grown used to barnowl's excellent support for composing long messages. As a compromise between allowing multiline IRC messages and not causing autowrap to send them inadvertently, send one IRC-level message (PRIVMSG or CTCP ACTION) for each two-newlines-separated paragraph in the user input to :irc-msg. Reviewed-by: Geoffrey Thomas <geofft@mit.edu>
Feb 5, 2010:
- 6:18 PM Ticket #129 (Appropriate handling of multiple AIM accounts) created by
- ":aimlogout" logs out all currently logged in AIM accounts, but what …
Feb 3, 2010:
- 5:05 PM Ticket #128 (No way to delete completions) created by
- If I accidentally typo a class name (or anything else that gets …
Feb 2, 2010:
- 11:48 AM Ticket #127 (context bug with :eperl) created by
- Using my DevelUtils plugin, run :eperl, and type something that has …
Feb 1, 2010:
- 2:20 PM Changeset [6249a76d]release-1.10release-1.7release-1.8release-1.9 by
- Restore lost handle_message hook. This was accidentally deleted in commit 36546fa4ddc4de9c607b0f7bb76f8041360cf9f1 when we removed the mainloop hook.
Jan 30, 2010:
- 9:26 PM Changeset [c3a47c9]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Move cachedmsgid from owl_filter to owl_view. This fixes a use-after-free bug: owl_function_create_filter sets inuse=1, removes the current filter (hence freeing it), then calls owl_function_change_currentview_filter → owl_view_save_curmsgid → owl_filter_set_cachedmsgid, which writes to the filter that was just freed. This means that you can no longer keep two independent positions in two different filters by repeatedly switching between filter1 → empty → filter2 → empty → filter1. But I doubt anyone even knew that was possible before. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 7:07 PM Changeset [bdb7c26]release-1.10release-1.7release-1.8release-1.9 by
- Only poll after posting if we're configured to.
Jan 25, 2010:
- 9:51 PM Ticket #126 (Allow the current message to not be in the current filter) created by
- Currently when you switch to a filter that doesn’t include the current …
Jan 23, 2010:
- 10:49 AM Changeset [fde6d80]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- owl_style_create_perl: Change refcounting convention. Make owl_style_create_perl assume ownership of an existing ref, rather than SvREFCNT_inc'ing its argument, for consistency with both perl APIs and other code paths that retain SVs.
Jan 19, 2010:
- 10:53 PM Changeset [2fec14b]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- perlglue.xs: Copy passed-in SV*s before storing them. Perl gives us a pointer to the same SV* that the caller was using, so if the caller mutates it in some way after the call, it affects our saved copy. This is definitely not what we want. Reported-by: William Throwe <wthrowe@MIT.EDU>
Jan 17, 2010:
Note: See TracTimeline
for information about the timeline view.