samjam: there you go. Please update and rebuild your entire tree
(libcitadel, citadel, and webcit) in order to get all the changes -- since
I made an API change you're going to need to refresh everything.
Both the OPNA and DLAT commands now accept either a part number or a
BTW I'd also like inline images to be able to refer to room-files as
well as
message attachments. It would be an advantage to a wiki/blog mode. Might
be
nice if it could also reference room-files off a different room on the
same
floor, etc...
Careful about that -- it won't
Does Berkeley DB even allow a partial fetch of a record? Even if it does
you're going to have to modify every single cdb_read() call in the entire
system.
I've got a bad feeling about this partial read idea. There could be a
*lot* of bad side effects.
Tonight I tried out the new webcit build with support for
multipart/related messages containing inline images. Seems to work very
nicely. I also tested a wiki room to make sure links still work there,
and they're still good too.
Nice work. This is a great feature that I've
One recurrence feature I need is the 5th day-of-week of month.
Ok, maybe we can do that. It would be really helpful if you could
generate such an event using some third party program and then post the
resulting iCalendar item to this room. That way I can import it into my
development
dothebart: the site-wide configuration screens in webcit appear to be
broken -- most of the controls are missing, and the ones which remain
contain text that says things like pre Handler [SERV:CFG] (in
Something to do with the templates?
dothebart: the above is a one line patch -- is it possible for you to roll
that into whatever package paulomorales is using?
did you know that theres web-developer or one of these other cool things
arround as js library so you can use it in IE too?
All I found was something called Microsoft Script Editor, and I can't
install it because the download requires Windows Genuine Advantage - which
doesn't run
I'll look at headers.patch tomorrow or over the weekend if I can.
Thanks.
1. alter citadel to try and use file buffers more (horrid job)
Nope. We put a lot of energy into eliminating temp files wherever
possible.
Couldn't you just swap to a page file, if your vhoster didn't give you a
swap partition?
W00T!! HTML Tidy running inside Firefox. A few cleanups fixed the
Internet Exploder Error 80004004 but I'm not telling you where the
problem is.
I'm running this patch on my systems and it makes pocket outlook work
again.
Would you mind maintaining that patch until after the next major release?
It's a significant enough change that I fear it could prolong QA for a
while.
I do want to start getting things cleaned up for
We've got to do something about these blank messages. Any idea what's
causing them?
Getting the following message in the Domain Names and Internet Mail
Configuration screen:
Handler [SERVCFG:FLUSHINETCFG] (in
'/appl/webcit/static//t/aide_inetconf.html' line 28); requires context of
type 3, have 0!
[SERVCFG:FLUSHINETCFG]
Here is a patch that enables you to select whether you want to reject an
e-mail based on SpamAssassin, or to do add X-Spam-Status and X-Spam-Flag
headers and accept. Please review and apply.
Ok, that patch looks good, and it's pretty much ready to apply, but can I
ask you to
Also, I want to increase the rules limit in webcit beyond 25, hope
no-one
minds?
That number was chosen arbitrarily. If you would like it to be more than
25, it's not a problem to increase it.
I waded around in that heap of for quite some time trying
to get
all sorts of things working in certain circumstances. I found one
definate
bug in libsieve that will probably never get fixed to do with address
matching. The interface into libsieve is a real horrible
Looks like libtidy will be required. There's some very nasty html
messages
going about, that libxml can't parse with it's friendly html parser :-)
Can libtidy run by itself, without attaching an XML parser?
I'm willing to let it be a compile-time option that you can turn on
Citadel unfolded the To: header into one enormous line and then
truncated it
at 1024 characters WITHOUT the trailing null and merged it into (thus
eating)
the Date: header, giving me a mixed header of 1061 characters.
Yikes, that *is* a problem. At the very least we should
I'm not too hot on the niceties of imap; but jow plausable is it to
graft a floor as if it were a persons own folder, so that f eatureless
imap clients like pocket outlook could still access all floors?
Here there be dragons. :)
IMAP is a very convoluted protocol, and
Even so, I really am not inclined to add all that weight to the Citadel
system. It's a great feature, but I don't think three new dependencies is
an acceptable price to pay for it.
libSieve isn't perfect but it was still easier than writing an entire
Sieve parser on our own. If someone wants to do that, then obviously it
would be welcome.
I really don't like the way it makes the comments look. IMHO it's *less*
readable than the plain language we use elsewhere.
*sigh*
This is tedious.
Eventually that should be part of the setup wizard type of thing we've
discussed before ... the code to run those screens is buried in there
somewhere but we never implemented it.
if you do if clauses, put the cheap conditions to the left, since the
computer evaluates it left to right. Stuff like IsEmptyStr() is real
cheap,
while strcmp isn't.
I never quite got to the point where I trusted short circuit evaluation.
Gotta work on that.
The funny thing is that I've been doing this in shell scripts for years.
I habitually write scripts that look like:
[ foo == bar ] do_something
..or even more frequently:
[ do something that returns an exit code ] || exit 1
Not even remotely interested in doing C++ and, since I'm not a
professional developer, there's little chance that I'll ever get roped
into it.
Oh yeah, and since I'm the lead developer of this project I can say,
we're sticking with C.
I did a little C++ years ago (remember Daphne,
Don't burn out!
http://www.freesoftwaremagazine.com/columns/supporting_your_free_software_do
nt_burn_out
It's a good read.
When I accidentally set the date to be 29th November (Saturday) repeating
every 5th Sunday then when editing it, the recurrance screen showed 5th
sunday but still selected 29th
Yes, that's known to happen. It really isn't finished yet. Initially I
focused on setting all the controls
I'm not impressed. They're only acting as an OpenID provider, not a
relying party. In other words, they make it easy for Microsoft to control
your identity, but they don't even make it possible to log in to a
Microsoft web site using a third-party OpenID.
OpenID will only be useful
ajc, can I ask where you are getting your libical from please.
As you figured out, it's the libical build at
freeassociation.sourceforge.net. What you might not have picked up yet is
that we are the maintainers of that project as well. libical is used in
dozens of different calendar
Sam:
I'm afraid we can't accept this patch.
Inline images are something we definitely want, but your patch integrates
libtidy/libxslt/libxml into the system as well. As I mentioned before, we
are not willing to add all of those new external dependencies to the Citadel
system at this time.
If
Yeah... fair enough.(libtidy is NOT needed, just haven't taken it out)
Thanks for understanding ... definitely don't take it personally, we'd love
to have some of your code committed to the tree in the future.
As for libtidy ... can it be used without an xml parser?
I suppose you've
Ok, that should fix the issue that a few people have had from time to
time, where a corrupt address at the end of a list of recipients can cause
libsieve to crash.
It crashed even when I added a linebreak. I even tried folding the
headers down to about 72 columns, with the same results -- if the last
address is chopped, it crashes.
Nasty stack smash, too. Last time I worked on something like this with
Aaron he couldn't even get it to fail on the
Those of you who are into schadenfreude will take great pleasure in
observing the next few weeks of weeping and gnashing of teeth as I try to
make sense of time zones and recurring events.
Don't forget to test it with messages that have attachments, as well as
submessages with attachments. That's where it gets tricky.
dothebart: I'm noticing that you used the short datetime in the message
headers. It should be using the full datetime.
Someone's been busy :)
yowza, that was a scary one. Good thing the code hasn't been released, or
we would have had a lot of unhappy IMAP users.
ah, was this related to the partial fetch stuff? no connection to that
sieve
fix? right?
The bug was always there, but partial fetch may have caused it to show
more.
It's *very* frustrating. Fixing time zone issues will probably take
longer than it did to *write* the entire recurring events thing. It's
recurring ... it's even recurring in the correct time zone ... then when
we hit a DST boundary everything shifts by an hour.
I want the whole world
DST really needs to be done away with.
Not just DST ... time zones need to be done away with. The whole world
goes to UTC and you just deal with it. So what if your work day in New
York is 2pm to 10pm? You'll get used to it in a week or two and then no
one ever has to guess what
Ok, so do we need to increment the version number of citserver and then
make webcit depend on it?
Hmm, even after a rebuild of libcitadel and webcit, I'm getting a title
bar on the login screen that says:
Handler [CONTEXTSTR] (in
'/appl/develcit/trunk/webcit/static//t/beginbox.html' line 3); requires
context of type 12, have 0!
[CONTEXTSTR(X)]
I love the comments in some of these commits.
What, doesn't *everyone* use the phrase HAHAHA WH BOING BOING
BOING YOWZ0RS to denote that a particularly tough issue has been
resolved?
I seem to recall that the phrase fuck me gently with a chainsaw is
fairly common in the Linux kernel.
ok, I'd say were at a point, where uncensored probably could be updated
to
SVN head.
Except it can't because the calendar is still broken :(
I just did a clean build of libcitadel and webcit, but this problem seems
to persist:
* Mailbox view: didn't find Template [summary_header] 14 14
(instead of the mailbox)
* Calendar view: shows parts of the Tasks, Calendar, and Contacts view
.. and no data
* Notes:
Ok, that looks mostly better ... it's displaying the correct view now, but
all of the views (except for mailbox) are now empty. You have to click
around for a bit before they show any data.
*sigh*
I'm getting tired of this. It was so much work to get recurring events
done, and after recurring events support turned out to require a complete
overhaul of the way we handle timezones, there's just so much more work
getting everything changed across the entire calendar
PACKAGE MAINTAINERS: I recommend applying this patch to your packages.
Easy Install has been patched, and I may go for the source tarballs
next...
Ooh, that could be nice.
Been there, done that, it's always answered by some random schmuck who
boldly declares that no open source groupware servers can replace Exchange
because they don't have the exact feature set of Exchange, don't
interoperate with Outlook, blah blah blah.
flew.. either way its a timesaver. Its not finished yet.. the resize
grippy
hasn't been put back yet, nor has drag and drop - excuse me for being
naive,
but can someone tell me what drag and drop is supposed to do in
listview?
That's for moving messages between folders. If you
This is all great stuff, and it's going to really be a useful
application.
However I'm concerned that we're getting to a point where I don't really
understand how WebCit works anymore. I'd appreciate if the two of you
could extensively document the templating and json stuff on the
Ok, this is interesting ...
I'm trying to move my development work back over to splorph.xand.com so I
can follow the libical changes without affecting the three production
Citadel sites on hawthorne.citadel.org.
Granted, it's a very old Linux (Fedora Core 3) but the way WebCit is
Better now, but I see a different problem: on a fresh install, for a new
user, the preferences screen is missing a bunch of fields.
when heading out on that again... what would be the appropriate place to
save
a global config in?
The room name is Local System Configuration but don't talk to it
directly, use the handy API functions:
CONF GETSYS|application/x-citadel-foo-config
and
CONF
Whew! That was a long haul. But it's almost finished!
At this point I think I can say that if we were accepted into GSoC and
had some wetback college kid attempt to do this, he would have screwed up
a lot of things and the cleanup would be enormous.
Couple of minor things left to
Doesn't matter too much. SVN doesn't solve all of CVS's problems, but
it's been more stable for us.
Hey Matt, I just read branches/webcit-newsummaryview/WISHLIST and I like
what I see there. Your vision for the mailbox view and icon bar matches
very well with what I had in mind (but never seem to have the time to
implement).
What timeframe did you have in mind for the merge?
I personaly most allways prepend *; should we use the above syntax?
Meh ... it doesn't look like something we really need.
locate_user_vcard
doesn't use its parameter username though auth.c puts it in... Bug?
Wow, that's interesting. Ok, I guess it's time to remove the extraneous
parameters, because the caller has already done the important work.
/* Look for commonly-found probes of malware such as worms, viruses,
trojans, and Microsoft Office. */
Sometimes I even amuse myself. :)
Not even close. We're in feature freeze but the code is barely alpha
quality. There's quite a bit of cleanup to do.
Ok, I think that was it. Am I correct in understanding that Msg-MsgBody
is just an ordinary wc_mime_attachment that happens to point to the MSG4
top level? r6932 populates the 'msgnum' with the proper message number.
This fixes the problem I was seeing, but are there any possible side
I like the safe mode idea. What is its behavior? Does it go to the old
webcit semantics or does it actually implement a non-ajax view?
Ok, here are some problems I see so far:
* Iconbar - the background is gone
* Iconbar - customize this menu appears above the last button instead
of below it, and the font is too big (did the css for it disappear?)
* When I click on Mail the message Loading messages from
More:
* Tabbed dialogs such as the event editor and site-wide configuration now
show the div that's common to all tabs (such as the Save/Cancel buttons)
as an odd looking bottom tab instead of spanning the entire width of the
dialog.
* The Chat function is gone. Pushing the
Wow, that's a lot of fixes in one shot. Thanks for responding so quickly!
A couple of things I noticed after this last update:
* switch to room list sometimes doesn't work. I'm still trying to
figure out a set of reproducible steps and will post them if I find them.
* When
Do you see a request for /json_roomflr in the webcit console? Which browser
version are you using?
I don't see that request. Firefox 2.0.0.11 on Linux.
Regarding your first list of regressions - are you sure iconbar had a
background? looking at the CSS on uncensored, its just white
Upon
So it's a generic drag and drop API that we can use anywhere? That's
useful -- I'd like to update the calendar screens in the future to allow
dragging appointments around the grid the way some others are doing it.
So ... does anyone know why OpenID isn't offered as a login option
anymore, or should I go take a look?
I'm also noticing that if I click on any name on the Contacts screen (or
presumably in any address book room), it doesn't bring up that contact ...
instead of just displays the This address book is empty screen.
I'm wondering whether the problem I am experiencing is the same one. When
I click Add, change, delete user accounts my webcit crashes with the
following backtrace:
#0 0xb7f64424 in __kernel_vsyscall ()
#1 0xb7bfb880 in raise () from /lib/tls/i686/cmov/libc.so.6
#2 0xb7bfd248 in
Clever, but I wish I didn't have to resort to such a sleazy kludge. If
there had been any way to do this inside the program code I would have
built it as a new API call within libical. It is rather stupid that
neither POSIX nor GNU thought anyone would have any use for a library call
that
Hmm ... I'm not sure what changed, but the mailbox view is totally empty
now. Both the normal mode and the safe mode display the throbber for a
split second, then the whole box clears out.
Matt:
gcc -I/root/ctdl/trunk/webcit -g -O2 -Wall --pedantic -DHAVE_CONFIG_H -c
-D_REENTRANT msg_renderers.c
./wmsg_renderers.c: In function bConditional_ROOM_DISPLAY_MSGb:
msg_renderers.c:410: error: bIterateStructb undeclared (first use in this
function)
Hey, at the risk of becoming something you dont want to be, ever think
of adding a status or mood or any of that crap?
.Goto Social Networking 3.0
It's all part of the long term plan.
But first we have to fix what we've already got. WebCit is currently
riddled with
Speaking of which ... I just tried out the current WebCit code on
Internet Explorer 6. The icon bar is a repeating display of icons instead
of one per line, and the room list doesn't display at all.
Do we have any statistics on what browsers webcit users use - say on
uncensored?
Sure, here are the statistics:
Overall:
Internet Exploder:46.21%
Firefox: 41.15%
Safari:5.15%
Opera: 3.17%
Chrome:
I'm at linux.conf.au at the moment, and the guys that are doing this
presented yesterday. As noted on the page, they told me we don't have to
do
anything to support it, unless we wish to support mutli-user
conferencing
So ... they just encode each video frame as an instant
Sadly it does seem we have to continue supporting IE6 for a while yet.
And I think I've finally figured out why some people think Citadel's UI
is ugly. They're viewing it on Internet Exploder. It's gorgeous on
Firefox, but it looks like crap on IE. This is partially my fault; I
don't
No, its just an XMPP extension to set up and tear down peer-to-peer
video sessions, the actual video streaming happens between the clients
themselves using RTP and some NAT-busting logic (ICE), so, 'SIP for
XMPP' may
be a better analogy.
Clever. We'd end up seeing in the
so that they can do a bulk move or bulk delete without having to read
the message. (This means that clicking the checkbox will not select the
message).
Not likely to happen; the current implementation is designed to act more
like a local client -- ctrl-click to multi select,
Ok, we're in pretty good shape now. All font sizes are now specified in
percent relative to the top level. In the future we might even consider
adding some controls to adjust that top level font size.
I'm looking at WebCit in IE7 and it's much nicer than IE6. Given that
IE6 is only a
and 2. wasn't blog going to be a pulldown option?
Not this version. That's going to be one of the new features of Citadel
in 2009. First we've got to finish what we've already started.
I see that the problem of grey/white bars getting out of whack after
deleting an odd number of messages has been fixed as well. Thank you for
taking care of that.
I did notice that the search function in the mailbox view is now broken.
Based on some quick testing I'm fairly certain that even when you put in a
search term, WebCit is sending MSGS ALL|||1 instead of a MSGS
SEARCH|... command.
In other news, since we're now in feature freeze,
A little commentary on the above commit:
I noticed this while trying to repair the display of multipart/related
images, which don't seem to be working right now. It was requesting a URL
like /webcit/mimepart/1/image012@01c97be4.f32b7040/ because the part
number was being passed to
Ok, so multipart/related images are working now.
QUESTION: the links generated by multipart/related always contain the
'/webcit' prefix. Is this acceptable? I didn't remove them because I
didn't want to break things running in that mode. I don't have a system
running in that mode so I
all rooms where gone in all room listings, I could enter them with goto
hidden room, it would barf some errors at me, and after that it will be
normal back again.
I've narrowed this down to a problem with configuring zlib compression in
our build system. On an existing system,
A note to everyone:
Translators have begun their work on localizing strings for the new
WebCit version (one has already returned their completed .po file), so we
are declaring an absolute freeze on any text strings for the time being.
If we find anything that is obtusely
I've just been doing a git-svn clone, and seen so many commits go by
that I
wanted to say thankyou to the guys who made those commits and brought us
citadel.
It's a lot of work but seeing people put the software to use (often on
very large systems) makes it all worth it.
(I'm going to at least check in a warning message; we can add the special
shutoff switch if we're in agreement that this is the way to go.)
These two bugs should receive priority:
1. OpenID login screen
2. Display of a single contact in an address book view
Once these are fixed, I think we will be at a point where we can update
Uncensored to svn head and begin some heavy-load testing.
601 - 700 of 1678 matches
Mail list logo
