Interesting thoughts.
Please take care of your family first. We're happy to have you aboard and
are excited about the project, but family absolutely has to come first.
That having been said...
We'll take the approach of treating Windows+Cygwin simply as a really brain
damaged flavor of unix.
Thoughts on changing the Quote command in the text client to
replyQuoted ?
Does anyone *ever* use pull quoting for anything other than replying to
the message they are citing? The reason this is suddenly important is
because if we made this change, it would encourage the use of
You could handle that by doing reply-Quoted, then Hold the message and
go to the next room to resume it there.
Yeah, I should use the hold mechanism.
Ok, I've made the change; this message is a test of it.
Quote has now been replaced by replyQuoted. It may be a jolting
experience for some initially, but I think we will be better off for it.
While it is true that you now have to
..and an item-by-item response to DaveCode feature requests, as if that
ever mattered...
http://web.archive.org/web/20020815092737/http://uncensored.citadel.org/cita
del/dave.html
Here is a list of things which need to be done on Citadel!
(But it's a list from August 2000)
http://web.archive.org/web/2830024719/uncensored.citadel.org/citadel/sta
tus.html
The entire revision history should all be there, even from before we were
running Subversion (the old CVS history was converted to Subversion
format).
Is there a way to obtain a stack backtrace in order to zero in on the
line of code that is causing these errors?
Also, keep
Config disappeared, or it just isn't pulling in POP3 anymore?
In other news, I just wasted the better part of the day (starting at about
09:00, and it's now 02:00 the next day here in Mt. Kisco, NY USA) tracking
down a libical bug.
I'm ready to kill myself right now because I probably fixed it during the
first hour or two of troubleshooting and
So I've committed this for peer review. ITS NOT TESTED AT ALL!!!
No kidding. It doesn't even compile :)
Manish: this is the developers mailing list. Please do not post feature
requests here.
Question is how did libtool get that path. Is it generated by libtool on
the
fly (need a newer libtool) or was it generated as part of the build
system
(bug in configure script or something)?
I don't know but I've got the same problem now, on at least one system.
This of
icaltime_compare_date_only_tz (new mozilla style syntax with a
timezone)
There is already icaltime_compare() and icaltime_compare_with_zone() so
you probably should call your new function
icaltime_compare_date_only_with_zone() in order to stick with the existing
naming convention.
I attempted to install this rev on Uncensored tonight, but it could not
resolve the symbol 'GetCount'.
So of course I noticed that SourceForge now accepts OpenID authentication,
and asked myself, how the heck are they handling all the non-web
services? You might recall that this was the stumbling block we
encountered when evaluating OpenID as a potential authentication service
for Citadel.
Thanks for the insight. Recurrence is defined in an iCalendar event as a
set of rules attached to the base event. libical contains a set of
functions to expand the recurrence. The way I see it, we have to do this
in three places:
1. Display code - expand recurrences when rendering a
AGH
It's *perfect* in Firefox but is completely broken in Internet Exploder.
Ok, so where are we with the current build on FreeBSD, Mac OS (matt), and
Windows (wduchene) ??
I suspect I owe some testing/bugfix to someone but I've lost track of
what's what.
davew:
Ever since your upgrade, your node has been mirroring network messages
back to uncnsrd.
This is the same problem we've had with the Citadel node that fleeb
operates, but we were never quite able to figure out.
Can you tell us what has changed on your system
Yikes. Even if that doesn't fix Matt's problem, it's still the right
thing to do.
(Don't get too excited. It's going to be a back burner project.)
The address book popup seems to have contacts sorted backwards (starting with
Z and ending with A). Where is the sort performed, so we can reverse that?
Wow. I think it's been longer than that, actually. I've been looking for
a software problem the whole time, when it was really just a configuration
problem. Gotta figure out a way to check for that condition.
SortByHashKey(List, 0);
is the sort thing; switching it to 1 would reverse the sort order.
Thanks, that did the trick.
The solution would be to have each node validate the nodename that the
caller
thinks it has connected to. In this case my node does NETP
uncensored|secret.
I was thinking about something like that too. Definitely the right way
to go.
Also discovered that Yahoo's implementation of OpenID only supports OpenID
2.0, which eliminates all that pesky simplicity. Instead it requires
three different service discovery modes, several of which require all
sorts of messy XML parsing.
Grr. Maybe I'll put OpenID on the back
Could it be fixed by upgrading prototype? Or are we already on the latest
version?
Regarding ECC crypto ...
You're not actually including portions of an ECC implementation in
Citadel, are you? You're just enabling those functions if they are
present in OpenSSL. Unless I am misunderstanding?
In that case it would seem that the burden of unencumbrance falls
Very happy to hear that it's working on Cygwin now. Soon, even the great
unwashed masses of Windows users will be flocking to the Citadel
revolution!! grin
Seriously though, we should get together sometime to sync up on a
vision for the Citadel system. It's been a while since we've
The plan is to write a script that checks subversion for new code, grabs
the
current head and then builds it something like easyinstall does now.
Before we go reinventing the wheel here, has anyone looked at Tinderbox?
http://www.mozilla.org/tinderbox.html
I tried using Microsoft's stupid Unix Compatibility Layer' thing,
Services For Unix (SFU, also known as STFU) ?
That's useful for only one thing: NFS/NIS client on Windows without the
need for Samba. Its use as a programming compatibility layer is just
plain awful. Cygwin can
Just use buildbot, so much easier than rolling another one.
Buildbot looks nice too. I'm going to assume that it's easier to
configure than Tinderbox, mainly because Tinderbox is a Mozilla product,
and knowing what a headache Bugzilla has been...
Anyway, it appears that most of
Bryon: please stay in touch and keep us up to date. The project sounds
interesting, and of course from a PR point of view it's great to have a
win against Zimbra. We will of course provide whatever server-side
support you require.
I don't think Queasy Install is available at all right now.
Today I did some troubleshooting for a customer who happened to be
running Citadel. The trouble report was mail is no longer forwarding
which to someone who knows the system a little better would actually
translate to my Sieve
So ... from what I'm reading here ... it sounds like y'all want me to
reinstate Queasy Install?
I was right! I was right!
But you were wrong about something else -- specifically, the bit about
your contributions being meager. There are no small contributions.
Everyone brings something to the party.
I'm in favour of adjusting the message timestamp to keep Webcit quick
and
also because we would split a message containing more than one VEVENT
into
Actually, we already do this. :)
Check out citadel/modules/calendar/serv_calendar.c on line 2009. During
the beforesave
Queasy Install has been reinstated, and is carrying the latest code as of
today.
There's a change that's long overdue :)
In the future I might even make the user number the primary key, so stick
closely to the provided API's and we should be fine...
Ok, there it is. Build packages from r6347 please.
Oops. r6348, not r6347 ... otherwise buildpackages will use the wrong
version number for webcit.
It was just great having that one glorious moment the first time I put in
an OpenID and clicked Login and it redirected through to the site
without having to enter local credentials.
Now, I just need to...
* Add the proper registration flow for a new user who signs in with
That looks pretty cool. Does it interface with a non-Launchpad SVN repo?
I wasn't able to find an answer to that question.
I've learned quite a bit about it during the (almost finished!)
implementation of OpenID 1.1 Relying Party support for Citadel. But this is
not the place for that discussion; let's take it to one of the tech geek
rooms if you want to go over what it is and what it can do.
It happened to me on June 2, on my test system. I was testing
export/import, so I wiped out all my data before importing. On the first
run it always crashed.
You might be able to tweak the Printable view to do something like that.
Our initial implementation of OpenID 1.1 Relying Party support is
complete! At this time I would like to have as many people as possible
test the code. In order to do so, please take the following steps:
1. If you do not already have an OpenID, please obtain one. You can go
to
The current Easy Install was ported over from Queasy Install.
What's the target platform?
Simplified HTML might be better accomplished using a light style sheet.
I don't know what would be oinvolved in making that happen, though.
I'm getting myself mentally prepared for implementing recurring events,
with an eye towards eventually doing CalDAV as well. One of the things
I've been thinking about has been that the way we render calendars in a
two phase operation, where we load up the calendar into memory from the
Thanks, matt ... that's a nice workaround to have for RSS until we have
true anonymous access mode available.
Yeah, it'll be pretty cool. Here's the MRTG config to make use of it.
You want to run this from a cron job that calls it every hour or so:
WorkDir: /export/home/httpd/mrtg
Options[_]: bits,growright
Target[citadel.example.com_accounts]: `( echo mrtg accounts ; echo
quit ) | nc
From my reading a phone entry like this:
PHONE;WORK;xxx xxx xxx
is as acceptable as a phone entry like this
PHONE;TYPE=WORK;xxx xxx xxx
but its not all that clear to me.
I don't like the look of that. I'm not sure if it's legal, but I'd be
willing to
since libical also has a vcard parser (meanwhile mrged back from
evolution...
;-)
should't we just start to use that? then we'd have a standard
implementation
When did libical get a vcard parser?
In any case, I really don't think I'm willing to spend the time
Ok, I see ... you're looking at libicalvcal. That's not a vCard
implementation specifically, but rather a generic API for handling data
formats that look like
BEGIN:VFOO
BAR;TYPE=BAZ:QUUX
EEK:The test test test is here
HAHA:More stuff here
END:VFOO
Should I have
OK, I can cope with that. Can somone ask the Binari (Bynari) guys how it
might affect them if our vcards suddenly start containing lines like
I'll email you the contact info for the person you need to speak to about
that. (Probably not a good idea to drop his name on this list.)
By the way, I'm on vacation this week ... sitting on a beach blissfully
avoiding work. I am online, but I'm only really spending time on the
computer in the late evenings, and my bandwidth is very limited because
I'm using a crappy cellular modem connection. It's very reliable but it's
I've upgraded bugzilla.citadel.org to Bugzilla 3.0.4 (the latest stable
version -- I'm not really inclined to be on the bleeding edge here). So
far I'm not seeing much of a difference.
Well, you saw how long it took for me to wrap my head around the hash
table idea. I eventually figured out that it's really useful. So you can
expect me to figure out the string buffer sometime in 2009. :)
I think the code should be:
if ( (!strcasecmp(type, directory))
(!strcasecmp(fqdn, host)))
return(hostalias_directory);
if ( (!strcasecmp(type, masqdomain))
(!strcasecmp(fqdn, host)))
(Don't get excited. We just detect it -- we don't do anything with it.)
That's great ... probably the way it should have been in the first place.
We could add a read only flag to the CtdlFetchMessage and a reference
count.
Each CtdlFetchMessage that is read only increases the ref count, each
CtdlFreeMessage reduces it. That way numerous read operations would only
need
one copy of the message in memory at a time. At the moment
Y'know, there's probably an easier way to speed up the import/export.
Strictly speaking, there's no reason why base64 has to be there at all.
The message format on disk is binary, BUT IT'S NOT ARCHITECTURE-DEPENDENT
BINARY.
So instead of ...
message
msgnum
reference
Attached is a patch which implements the above proposal. I haven't committed
it because I have not tested it. It compiles, and I believe it will work,
but someone needs to test it. (dave?)--- OLDmodules/vandelay/serv_vandelay.c 2008-07-17 23:43:13.0 -0400
+++
Doesn't UTF-16 require all new API's to handle strings with double-byte
characters? No thanks.
Go for it -- just be sure to wrap the whole thing inside #ifdef
TECH_PREVIEW so it doesn't get built into production Citadels until it's
ready.
but, now this line can grow to unlimited size... How'd the server handle
this? does it cope with it? will it just cut? is there a cheap acurate
like
the '-' in instant messages too? Or do we have SMTP alike folding?
Nope ... no such animal :(
so if we want to be able to handle this situation, we need to alter
citserver
to get a solution for this? so, implement a '-' version here?
I've thought about it. The problem is that ENT0 already uses its
SEND_LISTING input as the message text, so that path is not available for
a
Not *intended* but I completely understand why that would happen.
We used to have a command to send a file over the network, but I'm pretty
sure it's been removed by now.
The CtdlWriteObject() API no longer uses a temp file
That's been bothering me for a long time. Finally got around to doing
something about it. Let me know if I broke anything.
stat() the templates when loading them, and keep track of the modification
time. stat() them again whenever they are requested. If the modification
time has changed since the last access, burn the cache and load it again.
Umm, yeah, that would work too ... you didn't like my idea of calling
stat() to check cache validity?
now you can put
foo?_(Deleted)
into a template, and Deleted will be looked up in our .po's and the
string
matching your locale will be printed:
TEH ROX0R!! Does create-pot.sh automatically scan the templates and add
localizable strings like it does with the C source,
Wow, you're really going to town with this template stuff. Before long
it'll be a full scripting language :)
Bryon: that stuff is stored as arbitrary config data in the My Citadel
Config room. It's certainly how we'd implement storage of XMPP S2S
subscriptions.
comments? suggestions?
Yeah, it's all great ... now put it in the wiki so we'll remember it when
we need it :)
Will one of those last couple of commits fix the Quote function in webcit?
A little clarification of the above:
Someone recently moved ical.h from /usr/include to /usr/include/libical
.. it's been a little while since that change, so Citadel needs to look
for it there.I realize that this may annoy some people but the path
change seems to be mandatory in
Umm, yeah ... at the moment, if you want to build Citadel from SVN you
have to build libical from SVN. This will be fixed shortly with a new
libical release.
Yup, that's it. Aside from the year 1601, which is obviously wrong, it
also does BYMONTH=0 which chokes libical.
While waiting for six months of refcount_adjustments backlog to process
last night, I took the time to write a technical article about how THE
DREADED AUTO-PURGER does its thing.
http://www.citadel.org/doku.php/documentation:developers:deferred_processing
This is a pretty good
Why? What does nginx do that Apache wasn't doing?
plus not take a truckload of memory from the start; nginx doesn't show
up in
top M, while apache filled the upper half
Ok, I guess I'll try it on http://uncnsrd.mt-kisco.ny.us, which only
serves up a placeholder page, plus software upgrades for my phones.
Have you checked yet to see whether create-pot.sh is scanning those
templates for strings to be localized?
Since all of the unstable work is being done in webcit right now, should
we release new versions of libcitadel and the Citadel server? It's
probably time to get something out there.
We did, in fact, run it that way for a little while, because I had to
revert webcit to older code at one point.
Why did you feel that you needed to run FSCK?
Ok, we've got libical 0.33, libcitadel 7.38, and citadel 7.38 on the wire.
dothebart: do you want to build debs before we make an official
announcement?
I probably should have called that command something else. There is
seldom any reason to run it.
I'm assuming, based on my reading of the code and its output, that it
parses
the entire database and updates the reference counts system wide. I'm
assuming it is correct to run that after restoring from backups taken on
a
running database (i.e. restoring database and logs
I chatted with a developer on #openmapi a week or two ago. Sounds
promising but they have a long way to go. When they produce something
usable we can look into writing a Citadel back end for it.
I don't think we can integrate their stuff directly into Citadel --
there's too much Samba
** GRUMBLE GRUMBLE GRUMBLE **
I spent hours rewriting and debugging the routines that map a sequence
set to an array, twiddle the specified array members, then convert back to
a sequence set.
Around 2:00am I finally figured out that the logic was fine ... the
problem was that
I'm getting all sorts of errors building webcit, primarily in
cookie_conversion.c.
dothebart: you said this was building ok on your system, so ... if it's
still not generating any errors over there, could you try building it in
your home directory on hawthorne.citadel.org? I've got
samjam: welcome :)
Let's get up to speed. First of all, the place you want to be looking
isn't the MSG2 command, which is only used in the client protocol. You
have to go deeper, into functions such as CtdlOutputMsg() and
CtdlOutputPreloadedMsg() which then tell the back end to
The Other headers can be stored seperate, like the body is (but not in
the
same place), so that they can be fetched without fetching the body. (I
don't
like that idea)
Hmm ... how about this: we can store a special Citadel top-level header
that contains all of the headers
i.e. a reference to the Content-ID of one of the mime parts, and
replace
this with
src=/mimepart/message-id/attachment-no/content-disposition-filename
Yes, that's a great thing to do. I was planning on doing that someday
but if you want to implement it, great.
I
Were you thinking of writing that helper any time soon? :-)
How about I update citserver to offer a way to fetch attachments by ID
instead of part number, and you modify webcit to make use of it?
Sounds brilliant to me. I'll start work as if that were done.
Ok, I'll get to work on that.
One note about webcit: if you're going to make changes to the way we
treat URL's, you will want to be absolutely sure that you don't break
things when running in Wiki mode. We handle
501 - 600 of 1678 matches
Mail list logo