Timeline
May 16, 2010:
- 11:58 AM FAQ created by
- Move the "FAQ" over from the MediaWiki
- 11:49 AM Changeset [c45e1eb]release-1.10release-1.7release-1.8release-1.9 by
- Support 'make EXTRA_CFLAGS=blah' Signed-off-by: Nelson Elhage <nelhage@mit.edu>
- 1:45 AM Changeset [2cfc6d7]release-1.10release-1.7release-1.8release-1.9 by
- Punt dead code paths and pointless bad-enum cases Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [20aced3]release-1.10release-1.7release-1.8release-1.9 by
- Use GQueue instead of owl_list for the incoming message queue Otherwise, we process the queue in O(N^2) time. This isn't actually a bottleneck, but it becomes one whenever we create fake messages in bulk for testing purposes. Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [c314f39]release-1.10release-1.7release-1.8release-1.9 by
- We intern strings, so use this to punt the strcmp Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [3f6555d]release-1.10release-1.7release-1.8release-1.9 by
- Punt obarray and friends in favor of g_intern_string It's somewhat faster (about 1.5x by completely unrigorous tests), and it's less code for us to maintain. Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [f034ac0]release-1.10release-1.7release-1.8release-1.9 by
- Actually initialize owl_global in tester.c Fixes tests relying on variables being initialized. Better not to keep duplicating all the owl_global code. We tell ncurses to open a fake screen a /dev/null and run with it. Explicitly pass "xterm" as the terminal type to reduce reliance on the running environment. (Passing NULL defaults to getenv("TERM").) Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [3687413]release-1.10release-1.7release-1.8release-1.9 by
- Only enable pseudologin timer when we use it The feature defaults to off, and I imagine most people don't actually use it. In that case, we shouldn't bother waking up every 2 minutes to support it. (owl_zephyr_buddycheck_timer was moved to zephyr.c so it'd get a prototype. Also, it's zephyr-specific.) Signed-off-by: David Benjamin <davidben@mit.edu>
- 1:15 AM Changeset [7ba9e0de]release-1.10release-1.7release-1.8release-1.9 by
- Define constants for output switches Remembering 0, 1, and 2 is really just awful. Signed-off-by: David Benjamin <davidben@mit.edu>
May 15, 2010:
- 10:04 PM Changeset [700dc91]release-1.10release-1.7release-1.8release-1.9 by
- Add a start-command example to :bindkey since a lot of people asked about it
- 10:04 PM Changeset [9ed1278]release-1.10release-1.7release-1.8release-1.9 by
- Fix small :help bugs for bindkey and unbindkey
May 13, 2010:
- 5:38 PM const edited by
- Clarify which void * conversions work in C++. (diff)
- 5:26 PM const edited by
- State explicitly that foo ** may be implicitly converted to `foo … (diff)
- 5:21 PM const edited by
- Add a note about C library functions accepting pointers to strings. (diff)
- 4:57 PM const edited by
- Try to clarify the table by distinguishing between “changing” a char … (diff)
- 4:47 PM const edited by
- Add some examples to make clear that const int is equivalent to `int … (diff)
- 3:26 PM Changeset [447435a]release-1.10release-1.7release-1.8release-1.9 by
- owl_command_smartnarrow: Correctly use const. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 3:03 PM Changeset [ff426f9]release-1.10release-1.7release-1.8release-1.9 by
- owl_function_classinstfilt: Don’t leak class and instance on early return. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 3:02 PM Changeset [e54a746]release-1.10release-1.7release-1.8release-1.9 by
- owl_command_smartnarrow: Correctly initalize optind. If we don't reset optind, the first time someone runs smartnarrow, we'll start scanning arguments wherever the getopt() at startup left off. Signed-off-by: Nelson Elhage <nelhage@mit.edu>
- 2:28 PM Changeset [88e425f]release-1.10release-1.7release-1.8release-1.9 by
- Free temporary argv in done segment Otherwise we leak memory when failing out of the getopt loop. We also can punt the strdups, as getopt doesn't mutate them.
May 11, 2010:
- 7:42 PM Changeset [0881cdd]release-1.10release-1.7release-1.8release-1.9 by
- Use a separate pad for input wgetch calls a wrefresh on regular windows. I imagine this is for echo(), which we do not use. Instead, we make a pad and only ever wgetch on it. This also means that the various nodelay settings on other windows are unnecessary.
- 7:42 PM Changeset [bcff94d]release-1.10release-1.7release-1.8release-1.9 by
- Drop pointless doupdate before the event loop
- 6:46 PM Ticket #137 (Barnowl improperly wraps particular message) closed by
- duplicate: After further testing, davidben and I discovered this issue was due to …
May 10, 2010:
- 5:17 PM Ticket #137 (Barnowl improperly wraps particular message) created by
- In gdb's Barnowl 1.6 instance, there is (at least) one message that …
- 10:40 AM Changeset [05cfc78]release-1.10release-1.7release-1.8release-1.9 by
- Update Makefile.PL to refer to Net::Twitter::Lite
May 8, 2010:
- 6:31 PM Changeset [ab28a06]release-1.10release-1.7release-1.8release-1.9 by
- Don't complain about a missing config file.
- 1:43 PM WikiStart edited by
- (diff)
- 1:37 PM WikiStart edited by
- (diff)
- 1:37 PM release-notes/1.6 edited by
- (diff)
- 1:36 PM release-notes/1.6 created by
- 1:34 PM zcrypt created by
- 12:25 PM Changeset [5f08dbe]release-1.10release-1.7release-1.8release-1.9 by
- releasing-barnowl.txt: Convert to org-mode
- 12:25 PM Changeset [f536399]release-1.10release-1.7release-1.8release-1.9 by
- Update releasing-barnowl for the actual release.
- 12:25 PM Changeset [c62c755]release-1.10release-1.7release-1.8release-1.9 by
- do-release: Update the trac release text
- 12:10 PM release-notes created by
- 12:09 PM barnowl-1.5.1-announce edited by
- (diff)
- 12:09 PM release-notes/1.5.1 created by
- 12:08 PM release-notes/1.5 created by
- 12:08 PM release-notes/1.4 created by
- 12:07 PM release-notes/1.1.1 created by
- 12:06 PM barnowl-1.3-announce edited by
- (diff)
- 12:05 PM release-notes/1.3 created by
- 12:04 PM barnowl-1.0.5-announce edited by
- (diff)
- 12:04 PM barnowl-1.0.5-announce edited by
- (diff)
- 12:01 PM release-notes/1.0.5 created by
- 11:59 AM Changeset [adbf00a]release-1.6 by
- BarnOwl 1.6
- 11:55 AM WikiStart edited by
- (diff)
- 11:52 AM UsingPlugins created by
- 11:48 AM Plugins edited by
- (diff)
- 10:20 AM Changeset [161e879]release-1.6 by
- owl_function_zuserfilt: Don't leak the filter name.
- 10:19 AM Changeset [80ed757]release-1.6 by
- Always disable wrapping in single-line editwins. I also backported a hunk of d625cfd839f0b56d75199e576c0d3cab29a103b2for 1.6
- 10:18 AM Changeset [937ac795]release-1.6 by
- Set edit:maxwrapcols to 70 by default. New BarnOwl users may not be aware that sending wide zephyrs will make 80-column terminal users annoyed. Let’s enable 70-column wrapping by default, regardless of the sender’s terminal size. Users that know they need to send longer lines can easily disable it. Also, update the documentation for edit:maxwrapcols to suggest 70 instead of 60 as the recommended maximum. Signed-off-by: Anders Kaseorg <andersk@mit.edu>
- 10:17 AM Changeset [b96cc28]release-1.6 by
- Fix a bug in editwin redraw with echochar set.
- 9:40 AM Changeset [6cc3306]release-1.10release-1.7release-1.8release-1.9 by
- owl_function_zuserfilt: Don't leak the filter name. Signed-off-by: Nelson Elhage <nelhage@mit.edu> Reviewed-by: Anders Kaseorg <andersk@mit.edu>
May 7, 2010:
- 10:38 PM Keybindings edited by
- (diff)
- 9:40 PM Changeset [ecaec21]release-1.10release-1.7release-1.8release-1.9 by
- Make smartnarrow work better on CC:'d personals
- 9:40 PM Changeset [4727d31]release-1.10release-1.7release-1.8release-1.9 by
- Cache Zephyr CCs into an attribute on the message object.
- 9:40 PM Changeset [839697d]release-1.10release-1.7release-1.8release-1.9 by
- Make owl_message_get_cc_without_recipient return a GList. Signed-off-by: Nelson Elhage <nelhage@mit.edu>
- 9:40 PM Changeset [a41c8d1]release-1.10release-1.7release-1.8release-1.9 by
- owl_function_zuserfilt: Don't leak the filter name.
- 6:09 PM Ticket #132 (Reply does not handle newlines correctly) closed by
- fixed
- 6:08 PM Ticket #134 (perl extensions can confuse the editwin) closed by
- fixed
- 6:07 PM Ticket #125 (configure should check for more things) closed by
- fixed
May 6, 2010:
- 12:59 AM Changeset [c3031f3]release-1.10release-1.7release-1.8release-1.9 by
- Mask all handled signals while running pre-select actions. This closes the race condition where a signal could be delivered just before the pselect() call, which hangs until the next event. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 12:59 AM Changeset [1b5b975]release-1.10release-1.7release-1.8release-1.9 by
- Make terminal refreshes a pre-select action. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
May 4, 2010:
- 10:31 AM Ticket #136 (Ctrl-C during jabberlogin results in weird state) created by
- :jabberlogin takes a while, presumably because Jabber sucks. If I hit …
May 3, 2010:
- 9:09 PM Changeset [9d7a720]release-1.10release-1.7release-1.8release-1.9 by
- Always disable wrapping in single-line editwins.
- 9:09 PM Changeset [4d9e4254]release-1.10release-1.7release-1.8release-1.9 by
- Update the documentation for edit:max*cols.
- 9:09 PM Changeset [a4bbd80]release-1.10release-1.7release-1.8release-1.9 by
- Set edit:maxwrapcols to 70 by default. New BarnOwl users may not be aware that sending wide zephyrs will make 80-column terminal users annoyed. Let’s enable 70-column wrapping by default, regardless of the sender’s terminal size. Users that know they need to send longer lines can easily disable it. Also, update the documentation for edit:maxwrapcols to suggest 70 instead of 60 as the recommended maximum. Signed-off-by: Anders Kaseorg <andersk@mit.edu>
- 9:09 PM Changeset [d625cfd]release-1.10release-1.7release-1.8release-1.9 by
- Allow disabling of both automatic wrapping and filling. In addition, don't force wrapping at the window width unless the user specifically asked for it.
- 9:08 PM Changeset [7149832]release-1.10release-1.7release-1.8release-1.9 by
- Visually indicate soft line-breaks in the editwin.
- 8:10 PM Changeset [7ce30a9]release-1.10release-1.7release-1.8release-1.9 by
- Never draw into the rightmost column of the editwin.
- 8:00 PM Changeset [1af0a72]release-1.6 by
- Avoids segfault on NULL curswin for viewwin When the screen is too small, newwin() fails and gives back a NULL window. This avoids the crash.
- 8:00 PM Changeset [9e0bc3b]release-1.6 by
- Require resizeterm This allows us to get rid of the endwin, which causes problems. Specifically that ncurses will query the size itself on the next doupdate, causing a potential race. It also results in an annoying flicker that we see in typwindelta. If we punt passing arguments to owl_global_resize, I think we can support resizeterm-less curses with a bit of a dance: endwin() h,w = query_terminal_size() relayout_windows(h, w) doupdate() if (h,w != query_ncurses_size()) h,w = query_ncurses_size() relayout_windows(h, w) (This will only flicker given a race.) That said, this is annoying and we build with our own curses on athena 9. We may as well just assume it's there. Conflicts: global.c [nelhage@mit.edu: Backport to 1.6] Signed-off-by: Nelson Elhage <nelhage@mit.edu>
- 7:58 PM Changeset [b125668]release-1.6 by
- Fix an 'uninitialized' warning in recv:shift-right I have no idea why I didn't catch this when I was fixing recv_shift_left.
- 7:58 PM Changeset [774f439]release-1.6 by
- make viewperson, vp aliases for viewuser named in the same fashion as the default zlog directory for personals Signed-off-by: Kevin Riggle <kevinr@free-dissociation.com>
- 7:58 PM Changeset [7285db4]release-1.6 by
- Clear the resizepending flag before doing a resize, not after. This prevents us from losing resize events that occur during a resize operation. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
May 1, 2010:
- 11:35 AM Changeset [7141d75]release-1.10release-1.7release-1.8release-1.9 by
- Fix a bug in editwin redraw with echochar set.
Apr 30, 2010:
- 12:23 AM Changeset [fe4c786]release-1.10release-1.7release-1.8release-1.9 by
- Avoids segfault on NULL curswin for viewwin When the screen is too small, newwin() fails and gives back a NULL window. This avoids the crash.
Apr 29, 2010:
- 6:29 PM Ticket #135 (magic Discuss integration) created by
- I'm subbed to <discuss,*,%me%>. This means that when I get added to …
Apr 28, 2010:
- 2:27 PM Changeset [719173c4]release-1.10release-1.7release-1.8release-1.9 by
- Recreate the popwin on relayout This should fix redraws when resizing. We don't actually need to do it on every relayout, but I think that should wait until owl_window. The actual logic we want is probably something like parent-changed => relayout-children.
- 2:27 PM Changeset [06cc8d9]release-1.10release-1.7release-1.8release-1.9 by
- Simplify owl_popwin_is_active The current implementation is ridiculous.
- 2:27 PM Changeset [c21b8b7]release-1.10release-1.7release-1.8release-1.9 by
- Punt some redisplays in the popwin libpanel already takes care of stacking for us.
- 2:27 PM Changeset [644a0f8]release-1.10release-1.7release-1.8release-1.9 by
- Use touchwin in owl_function_full_redisplay redrawwin also memsets random chunks of the screen, which is completely not relevant for us.
- 2:27 PM Changeset [0e3cdf1]release-1.10release-1.7release-1.8release-1.9 by
- Punt a pointless debug and a trailing newline
- 2:27 PM Changeset [f92acd7]release-1.10release-1.7release-1.8release-1.9 by
- Only relayout on widget resizes We needn't compute a new window size and ask ncurses to resizeterm
- 2:27 PM Changeset [f7cf6c2]release-1.10release-1.7release-1.8release-1.9 by
- Allow scheduling of relayouts as well as resizes
- 2:27 PM Changeset [263320f]release-1.10release-1.7release-1.8release-1.9 by
- Require resizeterm This allows us to get rid of the endwin, which causes problems. Specifically that ncurses will query the size itself on the next doupdate, causing a potential race. It also results in an annoying flicker that we see in typwindelta. If we punt passing arguments to owl_global_resize, I think we can support resizeterm-less curses with a bit of a dance: endwin() h,w = query_terminal_size() relayout_windows(h, w) doupdate() if (h,w != query_ncurses_size()) h,w = query_ncurses_size() relayout_windows(h, w) (This will only flicker given a race.) That said, this is annoying and we build with our own curses on athena 9. We may as well just assume it's there.
- 2:27 PM Changeset [f9f88f3]release-1.10release-1.7release-1.8release-1.9 by
- Separate relayout from resize
- 2:27 PM Changeset [285bc9a]release-1.10release-1.7release-1.8release-1.9 by
- Refactor the ioctl and size work into a separate function
- 2:27 PM Changeset [1f3a423]release-1.10release-1.7release-1.8release-1.9 by
- Consistently use g->{lines,cols} for the final size We have various fallbacks and then ignore them, which will cause problems should we ever need the fallbacks.
- 11:11 AM Changeset [675a998]release-1.10release-1.7release-1.8release-1.9 by
- Fix an 'uninitialized' warning in recv:shift-right I have no idea why I didn't catch this when I was fixing recv_shift_left.
Apr 27, 2010:
- 11:46 PM Changeset [14965e5]release-1.10release-1.7release-1.8release-1.9 by
- make viewperson, vp aliases for viewuser named in the same fashion as the default zlog directory for personals Signed-off-by: Kevin Riggle <kevinr@free-dissociation.com>
- 11:14 PM Changeset [07f655b]release-1.10release-1.7release-1.8release-1.9 by
- Bump version for development.
- 5:21 PM Changeset [8479494]release-1.10release-1.7release-1.8release-1.9 by
- Clear the resizepending flag before doing a resize, not after. This prevents us from losing resize events that occur during a resize operation. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
- 5:21 PM Changeset [7892963]release-1.10release-1.7release-1.8release-1.9 by
- Mark variables modified from signal handlers as volatile sig_atomic_t. According to POSIX, this is the only type of variable (other than errno) that may be accessed from a signal handler with defined results. In practice, at least the volatile is important to prevent the compiler from reordering accesses. err_signal_info is a bigger struct that cannot be assigned atomically, but it’s only for debugging so we just do the best we can. Signed-off-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Nelson Elhage <nelhage@mit.edu>
Apr 26, 2010:
- 11:22 PM Changeset [72634c5]release-1.10release-1.7release-1.8release-1.9 by
- locker-update: Don't skip arches because the bare binary is missing.
- 7:52 PM Changeset [cef53f9]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- BarnOwl 1.6rc2
- 7:52 PM Changeset [a06334e]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Use a wrapper script for zcrypt Actual binaries get zcrypt.real pointed to them.
- 7:52 PM Changeset [0be790a]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Also create symlinks properly for zcrypt
- 7:52 PM Changeset [751ae8c]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- If the targets do not exist, remove the link This should correctly handle locker-updates to versions before and after the soname migration.
- 7:49 PM Changeset [0fd5bd5]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Incorporate the zephyr soname into the executable for *_deb50.
- 7:49 PM Changeset [a4b0572]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- locker-update: Create zephyr{3,4} symlinks as-needed Checks existence of target files before creating any
- 7:49 PM Changeset [0ee43c8]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Unify the beta and non-beta code paths
- 7:48 PM Changeset [60fcd71]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: des_ecb_encrypt doesn't NULL-terminate the output block. Output is a 'unsigned char [8]', so it couldn't even NULL-terminate it in place if it wanted to. So copy the decrypted buffer into a larger buffer that's guaranteed to have a NULL on the end.
- 3:46 PM Changeset [3001c11]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Add a test case for deleted syntax
- 3:46 PM Changeset [6dba228]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Fix filter tests to account for new deleted syntax
- 3:46 PM Changeset [ae333f0]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Add filterproc.c to BASE_SRCS Fixes linking of the two tester executables.
- 3:46 PM Changeset [4123da1]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Fix references to old editwin API in tester
Apr 25, 2010:
- 4:14 PM Changeset [1ea0249]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Add releasing-barnowl.txt
- 3:57 PM Changeset [1cfcab7]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Update copyright notices for 2010.
- 3:50 PM Changeset [858c9c4]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Our source actually expects DATADIR to be $(pkgdatadir).
Apr 24, 2010:
- 5:38 PM Changeset [5a57386]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Define DATADIR and BINDIR in the Makefile, not config.h This is the recommended way to do it, and deals with the fact that DATADIR and BINDIR themselves expand to variable references.
- 3:02 PM Changeset [9d2a105]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- BarnOwl 1.6rc1
- 2:44 PM Changeset [c18d069]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- add-changelog: Fix the usage message
- 2:22 PM Changeset [a1d98e4]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Support filtering on whether a message has been deleted.
Apr 16, 2010:
- 7:18 PM Changeset [63de71c]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- aim.h: Do the same stdbool.h dance This generates warnings (about redefining bool) on Solaris, but it compiles, and that's all I can bring myself to care about right now.
- 7:17 PM Changeset [13d6a53]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- scripts/locker-update: Install a zcrypt symlink
- 6:34 PM Changeset [75b0adb]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Check for stdbool.h in configure. Athena 9 Solaris doesn't have a working stdbool.h. The chain of defines and ifdefs is slightly modified from the "Particular Headers" section of autoconf's info pages.
- 6:04 PM Changeset [ef4700c]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Properly quote strings containing newlines or tabs.
- 6:04 PM Changeset [6e58ff2]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- Check for an unset mark in owl_editwin_replace_region. Fixes #134.
- 6:01 PM Changeset [7ba7d66]release-1.10release-1.6release-1.7release-1.8release-1.9 by
- zcrypt: Fix a segfault without any arguments.
Note: See TracTimeline
for information about the timeline view.