Timeline
Oct 28, 2011:
- 1:53 AM Ticket #205 (Segfault on :multi loopwrite; loopwrite; edit:cancel; edit:cancel) created by
- BarnOwl segfaults on `:multi loopwrite; loopwrite; edit:cancel; …[…]
Oct 17, 2011:
- 8:15 PM Changeset [a19c9a3] by
- Merge 389d48749c4eb68f218e9fd6faf953ba542bfb75 into ef4074b758108eb46e2b1884e950d6f092a01260
- 8:12 PM Changeset [389d487]release-1.10release-1.9 by
- Switch to interactive context before sourcing the startup file There are no invariants that don't hold before the previous place we flip the interactive bit; although default_style is applied afterwards, there is still a style set before that code. There is no code which reads the READCONFIG bit, so no behavior should change from there. Setting the interactive bit makes owl_function_load{login,}subs noisier, but that's fine. A number of commands were labeled OWL_CTX_INTERACTIVE, but that doesn't change much because, unless the command is OWLCMD_*_CTX, the bits are ignored. What does change is that .owl/startup is allowed to leave with a context on the stack. In particular, this /finally/ fixes #161. More fundamentally, this makes .owl/startup identical to the user having run those commands in succession, which is a nice simplification.
Oct 15, 2011:
- 8:17 PM Changeset [822e9bd] by
- Fix SASL digest-uri to be acceptable to ejabberd. This is, approximately, 68ad6bc and 8fe68e0 from the XML-Stream repo. The problem is that, if example.com pointed SRV records at foo.example.com, we'd send "xmpp/foo.example.com" (which is, arguably, what RFC 2831 calls for, but community consensus seems to disagree) rather than "xmpp/example.com". ejabberd would also accept "xmpp/foo.example.com/example.com", and that would follow RFC 2831 more closely, but I don't want to introduce divergence from the upstream unless there's an observed interoperability problem that would be fixed by it.
Oct 14, 2011:
- 12:09 AM Ticket #204 (zpunt command should be consistent with the subscribe command) created by
- :sub message * @cs.cmu.edu and :zpunt message * @cs.cmu.edu both …
Oct 13, 2011:
- 3:36 AM Ticket #203 (Unsafe strcat on overlapping strings in owl_zephyr_smartstripped_user) created by
- From owl_zephyr_smartstripped_user: […] This strcat call might …
Oct 11, 2011:
- 11:47 PM release-notes/1.8 edited by
- Gmail some spaces got eaten. Also random wording cleanups (diff)
- 3:31 PM WikiStart edited by
- BarnOwl 1.8 (diff)
- 3:26 PM release-notes/1.8 created by
- 2:05 PM Changeset [4f8535e]release-1.8 by
- BarnOwl 1.8
Oct 10, 2011:
- 8:07 PM Changeset [6503194] by
- Merge e21b9212952a5f6ff9f30193013153a0470944bb into ef4074b758108eb46e2b1884e950d6f092a01260
Oct 3, 2011:
- 12:33 AM Changeset [ef4074b]release-1.10release-1.9 by
- Cleanup in owl_zwrite_create on error cases Before, the caller was required to cleanup, even on error. This is kind of weird. In particular, owl_zwrite_create_and_send_from_line forgot to so there was a small leak in there. Also with 6329cd5136a27361dedf94bb92e20fa2d327e43c it's now possible to error from owl_zwrite_create_from_line before the owl_zwrite is initialized, which means that owl_zwrite_new_from_line on misquoted strings explodes. Oops. Although I don't think this is actually possible to trigger now. Also be consistent about whether rv != 0 or rv < 0 is the error condition. Barring any actual arguments one or another, I'm going to arbitrary declare it's the former since that function doesn't really have a useful number to return.
Oct 2, 2011:
- 3:38 PM Changeset [8da4e96] by
- Add command to get a buddy's status This could be used by a user directly, though it's probably more likely to be used by a module author of some sort.
- 8:22 AM Changeset [9579509] by
- Merge 215b8d5017196e5ddc3408e367c01570c41c8680 into f34728b01e16a1cadffac064a947fd75ff28581a
- 1:31 AM Changeset [4af1241] by
- Add completion for jabberlogout
Oct 1, 2011:
- 11:00 PM Changeset [f49afc3] by
- Merge f34728b01e16a1cadffac064a947fd75ff28581a into f89cc6fa54344f8f6b48601228af46e8fc95caea
- 10:59 PM Changeset [f34728b]release-1.10release-1.9 by
- Add completion for reload-module
- 9:07 PM Changeset [ba4d1ad] by
- Merge 923c3f6c4a953b7c17e7a90328dcf2451255bdd2 into f89cc6fa54344f8f6b48601228af46e8fc95caea
Sep 30, 2011:
- 11:47 PM Changeset [215b8d5] by
- Cleanup in owl_zwrite_create on error cases Before, the caller was required to cleanup, even on error. This is kind of weird. In particular, owl_zwrite_create_and_send_from_line forgot to so there was a small leak in there. Also with the previous set of commits it's now possible to error from owl_zwrite_create_from_line before the owl_zwrite is initialized, which means that owl_zwrite_new_from_line on misquoted strings explodes. Oops. Although I don't think this is actually possible to trigger. Also be consistent about whether rv != 0 or rv < 0 is the error condition. Barring any actual arguments one or another, I'm going to arbitrary declare it's the former since that function doesn't really have a useful number to return.
- 11:09 PM Changeset [f89cc6f]release-1.10release-1.9 by
- Use the pre-parsed argc/argv in building owl_zwrites When called programmatically, the buff argument is, unfortunately, bogus. Quoting that correctly is somewhat more of a pain due to skiptokens commands. (See e3c8332fa85642544dba1222912b77cf6e32ce8c and 6a7111370ea0e1de2c3eb86a85f34cd68848d4b1. Though we really should fix that properly sometime.) This fixes programmatically sending zwrites from perl with spaces and such. For instance, BarnOwl::zwrite('-c', 'barnowl', '-i', 'spaces yay?'); and BarnOwl::zwrite('davidben', '-m', "mangoes\nare\ntasty"); It also, as a side effect, brings us one small step closer to eliminating the buff argument from C commands. Reported-by: Alex Dehnert <adehnert@mit.edu>
- 11:09 PM Changeset [6329cd5]release-1.10release-1.9 by
- Allow building a zwrite from an argv Add new versions of owl_zwrite_create and owl_zwrite_new. Note that this means we rebuild z->zwriteline from scratch unconditionally. The only behavior change is that the user's spacing and quoting is not preserved in zwrite locktext. For instance :zwrite davidben -i "spaces yay?" now gets a locktext of ----> zwrite davidben -i 'spaces yay?' I think this isn't a big deal. It already happens for every other *write locktext, and if we kill the buff argument, it'll happen there too.
- 9:35 PM Changeset [a962f5c] by
- Merge 2dd9c4831cbda78650784ec9d10c3f6820bfdc01 into 78033261fc599d8d986e1c32b68523132b853d40
- 9:32 PM Changeset [2dd9c48] by
- Use the pre-parsed argc/argv in building owl_zwrites When called programmatically, the buff argument is, unfortunately, bogus. Quoting that correctly is somewhat more of a pain due to skiptokens commands. (See e3c8332fa85642544dba1222912b77cf6e32ce8c and 6a7111370ea0e1de2c3eb86a85f34cd68848d4b1. Though we really should fix that properly sometime.) This fixes programmatically sending zwrites from perl with spaces and such. For instance, BarnOwl::zwrite('-c', 'barnowl', '-i', 'spaces yay?'); and BarnOwl::zwrite('davidben', '-m', "mangoes\nare\ntasty"); It also, as a side effect, brings us one small step closer to eliminating the buff argument from C commands. Reported-by: Alex Dehnert <adehnert@mit.edu>
- 9:32 PM Changeset [d6c2e03] by
- Allow building a zwrite from an argv Add new versions of owl_zwrite_create and owl_zwrite_new. Note that this means we rebuild z->zwriteline from scratch unconditionally. The only behavior change is that the user's spacing and quoting is not preserved in zwrite locktext. For instance :zwrite davidben -i "spaces yay?" now gets a locktext of ----> zwrite davidben -i 'spaces yay?' I think this isn't a big deal. It already happens for every other *write locktext, and if we kill the buff argument, it'll happen there too.
- 9:32 PM Changeset [7b89e8c]release-1.10release-1.9 by
- Add owl_argv_quote convenience function Did we really never write this thing?
- 7:58 PM Changeset [d953ede]release-1.10release-1.9 by
- Rename owl_zwrite_new to owl_zwrite_new_from_line owl_zwrite_new will take an argv, like god intended.
- 7:24 PM Changeset [923c3f6] by
- Export zephyr_zcrypt function to Perl The currently available infrastructure doesn't appear to allow sending multi-line zcrypted zephyrs from Perl (or at least not reasonably).
- 8:23 AM Changeset [6df57d4] by
- Added a perl hook for :away Note: The sepbar no longer distinguishes between AWAY, A-AWAY, and Z-AWAY.
- 8:22 AM Changeset [57ad328] by
- Added support for calling perl subs with various return types
- 8:22 AM Changeset [0c71c58] by
- Refactor perl calls through a single method I don't know the perl/C interface well enough to figure out the best way to standardize the many variants that we use to call perl code. Perhaps we should standardize the error messages, and put less knobs on the boilerplate macro.
- 8:14 AM Changeset [e21b921] by
- Add tests for enum variables.
- 8:14 AM Changeset [cce9369] by
- Allowed custom validsettings for other non-boolean perl variables.
- 8:14 AM Changeset [b54b06a] by
- Added functionality for creating enum variables from perl
- 8:12 AM Changeset [c0e728a] by
- Added functionality to create new enum variables from C
- 8:12 AM Changeset [5f784ec] by
- Refactored perl variable creation code in preparation for ambiguous types When we allow enums in perl, we can't be sure that recreation of an int isn't actually changing it into an enum (though, really, variables should only be recreated when you reload module code). This makes BarnOwl reinitilize all the fields, but keep the current value, if there is one.
- 8:07 AM Changeset [7803326]release-1.10release-1.9 by
- editwin callback for canceling the editwin The code for editwin callbacks (called when the editwin is created) has been extended so that callbacks are called when the editwin is canceled. The old (perl) editwin callbacks still exist and have the same functionality that they always have. They are now deprecated.
- 8:06 AM Changeset [e89ec48]release-1.10release-1.9 by
- Made BarnOwl::start_* wrappers around a new BarnOwl::start_edit A new function, BarnOwl::start_edit, was created as the preferred interface for starting the edit_win from perl. The perl interface to the edit_win is backwards compatible. This is mostly preparation for the next commit, which generalizes the syntax of starting an edit_win to allow callbacks on cancel.
- 8:06 AM Changeset [c737503]release-1.10release-1.9 by
- Improved consistency between owl_function_start_* functions
Sep 28, 2011:
- 5:31 PM 256Colors edited by
- Emacs bug is fixed in Emacs 24 (diff)
- 5:19 PM 256Colors edited by
- The screen-256color underline bug was fixed in ncurses 5.8 (diff)
Note: See TracTimeline
for information about the timeline view.
