Timestamps are in UTC.
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4649 * RyanKing * (+6) Feed - cleared up the verbiage re: classnames
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4650 * RyanKing * (-39) Entry -
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4651 * RyanKing * (+6) Entry Title - cleaning up verbiage re: classname
[[using-utf-8]] N http://microformats.org/wiki/using-utf-8 * Tantek * (+2107) first draft
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4652 * RyanKing * (+13) Entry Title - changing verbiage re: class names
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4653 * RyanKing * (+1)
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4654 * Ant * (+15353) added Trillian & Google Talk formats
[[hatom]] M http://microformats.org/wiki?title=hatom&diff=0&oldid=4655 * RyanKing * (-1) Implementations - just some whitespace cleaning
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4656 * Ant * (-9511) revert to before I duplicated the page
does the wiki do merges?
no
who is Ant?
it looks like we're seeing a problem with chat-examples that has occured with other -examples pages
namely, that *formats* are being added, rather than discrete examples
uh yeah
not sure i totally get the difference, is there a good rundown on what's meant by that?
examples = html
formats = other stuff
cks, see http://microformats.org/wiki/examples
here is one good litmus test:
Can you paste in a URL which points to the actual chat example *content*?
If so, you have an example. If not, you're probably documenting somebody's format.
Nothing wrong with documenting a format, but do so in the proper place, in a *-formats page, not *-examples
will move the chatzilla one over until it's proven somebody actually publishes it
[[examples]] http://microformats.org/wiki?title=examples&diff=0&oldid=4657 * RyanKing * (+68) Current - added note about formats
the rdf ones get a little weird, they're definitely published, plenty or urls, but by the spirit of the rule seem more like a format
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4658 * ChristopherStJohn * (+31)
[[chat-formats]] N http://microformats.org/wiki/chat-formats * ChristopherStJohn * (+2003)
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4659 * ChristopherStJohn * (-1972)
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4660 * RyanKing * (-122) Nomenclature - cleaning up some language
cks, the idea behind -examples is to see emergent patterns of what people do on their own
people are certainly influenced by formats and will publish in formats, but we really want to now, "when left to their own devices, what do people publish?"
we can then analyze those examples for 'implied schemas'
sure, it gets a little weird for irc logs, though, since people don't actually ever create them without help from a program
there is no manual authoring
right, but what about conversations in general?
what about this: http://kitta.net/2006/02/01/i-am-heartbroken/
where she re-created the irc conversation in html
well, revisting the "why" is probably useful, but the original mailing list discussion in the archives (appeared-to) validate that irc logs in particular were their own problem space
[[examples]] M http://microformats.org/wiki?title=examples&diff=0&oldid=4661 * Tantek * (-3) Good Examples of Examples -
in any case, i'm moving the clearly-formats stuff over into a chat-formats page referenced off the examples page, i assume that's the correct thing to do?
yes, thanks
kingryan, perhaps add the kitta example to the chat-examples page
:D
I'll do it once cks is done moving stuff around
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4662 * ChristopherStJohn * (-4125)
[[chat-formats]] http://microformats.org/wiki?title=chat-formats&diff=0&oldid=4663 * ChristopherStJohn * (+4243)
so, TantekC, I'm looking at http://microformats.org/wiki/hatom, trying to see what can be improved
and I'm tempted to move some of it out into a faq
there's too much illustrative stuff buried deep in the page
i have to go grab a bite to eat, there's still some stuff that neads to be moved and i suspect i'll have a couple more questions, but i won't be returning to edit for about an hour
and I wish we could kill most of the opacity stuff
cool
kingryan, i think most of the opacity stuff is unnecessary
though i'd rather see us discuss it on the list and form a consensus rather than do any "executive editing"
let's pow-wow on it asap
k
i remember DavidJanes feeling pretty strongly about it
for now, I'm just trying to de-cruft the spec and clean it up
yes, definitely a good thing to do
I'm making sure the proper terms are used and such
and taking out speculative aside comments
[[hatom]] http://microformats.org/wiki?title=hatom&diff=0&oldid=4664 * RyanKing * (-314) Entry Permalink - removing some cruft
kingryan, based on the utility of sharing UTF-8 tips, specifically in the context of microformats and moving/parsing/processing microformatted data, i've added a using-utf-8 page
url?
[[examples]] M http://microformats.org/wiki?title=examples&diff=0&oldid=4665 * Tantek * (+57) moved "not formats" to "not" section under "Behavior"
http://microformats.org/wiki/using-utf-8
I figured perhaps you could add some of what you have discovered with UTF-8 handling with PHP/XSLT and Tidy.
including perhaps warnings of problems found/encountered
(in the MIddleware section - feel free to rename)
heh
I wodner how much cursing at perl and mysql I should add there
KevinMarks, as much as you like ;)
with hopefully helpful suggestions as to how to avoid the curse-inducing situations
avoid perl, avoid Java ;)
avoid PHP
poke MySQL wiht a large stick
avoid the web?
wow, json got mentioned in at TAG finding
[[hreview-faq]] http://microformats.org/wiki?title=hreview-faq&diff=0&oldid=4666 * Tantek * (+362) use tags for stocks also
about to start doing some editing on the chat-examples. one thing, given the previous conclusion that irc logs were their own thing, and the new suspicion that they are part of a more general problem, it's probably worth nailing down what exactly the problem is
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4667 * ChristopherStJohn * (-1019)
[[chat-formats]] http://microformats.org/wiki?title=chat-formats&diff=0&oldid=4668 * ChristopherStJohn * (+1019)
cks, yes its worth nailing down the problem
but don't ask me, 'cause I don't have any problems :D
i'll post where i'm coming from on the mailing list, see if people care enough to follow up...
[[chat-formats]] http://microformats.org/wiki?title=chat-formats&diff=0&oldid=4669 * ChristopherStJohn * (+2342)
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4670 * ChristopherStJohn * (-1174)
Enric is a media Software Developer and Videoblogger located at http://www.cirne.com
cks, yes, the clearer the problem statement you can provide the better.
[[using-utf-8]] M http://microformats.org/wiki?title=using-utf-8&diff=0&oldid=4671 * Tantek * (+136) HTML -
[[chat-formats]] http://microformats.org/wiki?title=chat-formats&diff=0&oldid=4672 * ChristopherStJohn * (+1461)
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4673 * ChristopherStJohn * (-1179)
[[chat-examples]] http://microformats.org/wiki?title=chat-examples&diff=0&oldid=4674 * ChristopherStJohn * (+239)
atamido: http://barcamp.org/BarCampAustin?
go barcamp, go
!
[[hatom-issues]] http://microformats.org/wiki?title=hatom-issues&diff=0&oldid=4675 * MarkRickerby * (+27) Alternatives -
Enric is a media Software Developer and Videoblogger located at http://www.cirne.com
cks: I don't drink, can I still have fun? :P
atamido: bar = foo++, not bar=drink
although, from what whurley was saying, you may have a point
[[to-do]] http://microformats.org/wiki?title=to-do&diff=0&oldid=4676 * MarkRickerby * (+257) New Person 2 -
[[Main Page]] http://microformats.org/wiki?title=Main_Page&diff=0&oldid=4677 * ChristopherStJohn * (+18)
[[start-simple]] http://microformats.org/wiki?title=start-simple&diff=0&oldid=4678 * MarkRickerby * (+1004) draft content
kingryan is ryan king
trovster is a web developer from the UK who writes on http://www.trovster.com and runs www.csslounge.co.uk
karlUshi is karlcow
bluesmoon is Philip from India & writes often on livejournal & sometimes about tech stuff on http://bluesmoon.blogspot.com & restaurants on http://bluesviews.blogspot.com & local food secrets on http://bluesfood.blogspot.com
[[hcard-brainstorming]] M http://microformats.org/wiki?title=hcard-brainstorming&diff=0&oldid=4679 * Atamido * (+30) Implied "FN and N" Optimization (proposal) - +1 NICKNAME
Atamido is Paul Bryson, http://orangeman.commo.de/
RobertBachmann is Robert Bachmann <http://rbach.priv.at/> and lives in Austria (Timezone: 01:00)
RobertBachman : doesn't look like your script format is up on chat-examples, i'll add it now unless you planned to...
isn't
I'm not sure if it makes sense to list it there ... as I'm the only one who uses it
mlinksva is Mike Linksvayer and from Creative Commons
hmm, you actually publish stuff already using it?
the baroque details of "THE PROCESS" sometimes confuse me, might need to bring in a referee :-)
hober is Edward O'Connor and works for EVDB on http://eventful.com/ and lives in San Diego, CA (-08:00)
cks: The format is used (AFAIK exclusively) by the logs for this IRC channel. http://rbach.priv.at/Microformats-IRC/2006-02-02 -> View Source
greetings
could someone verify that they can see this URL: http://www.w3.org/2006/03/01-TechPlenAgenda.html
TantekC: I can see it
check out the 10:30 session :)
indeed
very nice
[[events]] M http://microformats.org/wiki?title=events&diff=0&oldid=4680 * Tantek * (+90)
Yay, I can connect to the software now!
Wooo!
Oops, wrong channel. :P
atamido: care to share the joy?
A $50,000 hardware/software package from Motorola.
Turns out the default permissions they ship with don't allow any external connections, which is how you use the software.
Of course there is absolutely nothing to indicate this anywhere in the software or documentation.
heh
You have to look at the database itself.
Or use Ethereal and look at the contents of the packes it is sending back and forth.
blake is allegedly human. Blake, also known as Cortland M. Setlow, studies at swarthmore.edu and enjoys building things, exploring buildings, and physics. He currently sleeps during the day.
^ what I did.
?help
Actually, the documentation is basically worthless.
It is horribly incomplete and full of errors.
Apparently the software requires an SVGA soundcard. :P
*cough* your joy in actually getting it working is understandable :-)
Atamido: It seems every $50,000 package of hardware/software is an undocumented piece of crud. I'm suprised you could connect over ethernet and weren't forced to use GPIB :)
?def cks is Christopher St. John and has a brain filled with inane drivel from irc chat logs
cks is Christopher St. John and has a brain filled with inane drivel from irc chat logs
nice one tantek
no get them to mark it up with hcalendar
hi DanC, this page: http://www.w3.org/2006/03/01-TechPlenAgenda.html
at a technical level, yes. but I don't know what tool steve is using or whether any edits I made would be preserved
let alone whether the times are stable. (and I'm sure he wouldn't update title="YYYY-MM-DDTHH:MM:SS+0100" stuff)
to turn <span class="start">8am</span> to <span class="dtstart start" title="2006-03-01T08:00:00+0100">8am</span>
well, there is the tabular stuff too
TR/html4/loose.dtd . it's not even XML. I'd have to tidy it. I'd need to coordinate that closely.
tabulator stuff?
the way of using table head for columnar info
http://microformats.org/wiki/hcalendar-brainstorming#Tabular_Data
don't think the parser does that yet
DanC, "headers" attribute in HTML4
X2V supports it
oh, it does?
nice
it has since September!
live example:
i still haven't seen an example of using headers attributes that makes sense to me
http://we05.com/program.cfm
you should update the main hCalendar page to say so
Kevin, yes, that is on my hCalendar to-do list
very cool that you got it in
I don't see anything on program.cfm that calls for header attributes. each <tr> is a vevent, no?
DanC, e.g. this works: http://feeds.technorati.com/events/http://we05.com/program.cfm
no
location
some <tr>s have multiple events
some don't
ah... location
and time
oh and time on the oterh axis
ugh... more repeated data: <th axis="location" id="mcfarlane">Stream 2 (<span class="location">McFarlane</span>)</th>
ugh... and even more: <td class="vevent" colspan="2" headers="guthrie t1615">
how is this better than putting all the data in each cell? it's even *more* redundant than that
DanC, it is better because it preserved the visual layout that the page author created
those id's could have been id="L1", id="L2" instead of id="mcfarlane" etc.
same thing with t1615 - it could have simply been "slot4" or something abstract like that. just because it *looks* like data, don't assume that it is.
the actual data is only there *once*, in the respective <th>
TantekC: BTW, you broke that page
but L1 and L2 are still data that have to be maintained
your last name causes "escaping malformed URI reference" warnings
whihc page?
http://weblog.burningbird.net/2006/02/02/a-story-of-cane-and-able-and-the-browser-that-rode-chariots/
qid, which page?
the program.cfm you were just talking about
Kevin is that yet-another-burningbird-url-that-has-nothing-to-do-with-microformats or am I missing something?
if I read RFC 2396 correctly (and if it's still normative), "Çelik" would be a valid fragment identifier
qid, that page works just fine for me, in my browser(s), in iCal.app using the feeds.technorati.com link etc.
what application gave you an error?
Html Validator 0.7.7 firefox extension, and it's a warning, not an error
qid, W3C thinks it is fine: http://validator.w3.org/check?uri=http%3A%2F%2Fwe05.com%2Fprogram.cfm&charset=%28detect+automatically%29&doctype=Inline&verbose=1
maybe that's a problem with the firefox extension
apparently it claims "Çelik" as the fragment identifier is a malformed URI sequence, except I believe it is not according to the RFC
looks legit to me
then it is forgetting to do entity resolution on attribute values
however, the element they are trying to link to actually has an id of "celik", so the link is broken
that's a pretty big bug
well, it depends how it does it too
no it doesn't depend
you have to resolve the entities *first*
per HTML/SGML parsing rules
then you apply any attribute specific semantics
i.e. the fact that it is an href and thus a URL
if it does entity resolution into docunement encoding, then it can become awkward
the warning references RFC 1738, which says nothing about fragment identifiers, although it does claim that the & and ; characters are reserved
qid, it shouldn't be trying to parse the & and ; as part of the URL - that's its bug
ah, ok
by the time it is trying to parse it as a URL, the HTML parser is supposed to have already resolve the character entity into a single character
anyway, back on topic
DanC, what do you think of your chances of adding hCalendar to that page and having it "stick"?
not good, today, due to my own priorities and the volatility of the page. In a week or two, odds are better
I can imagine coralie might be interested to learn how to do it
TantekC: actually... I think that is in fact an error, as U+00C7 isn't one of the allowed characters in RFC 2396
it should have been URL-encoded, not turned into an entity
thats what I was getting at on order of resolution
in practice, lots of sites do have urls in doucment encoding
*document
but the %-escaped utf8 is a better way moving forward
I can't fathom why their page that claims to be UTF-8 doesn't just use UTF-8 characters, instead of turning them into entities
so perhaps they *are* resolving entities, and thus the warning is correct
if it is claiming that & and ; characters are reserved, then no, it is not resolving the entity
but Kevin is right, non-ASCII7 chars in URLs should be %escaped
(until we get to a version of HTML that uses IRIs)
it doesn't explicity say what it doesn't like
ah, ok, that was just your inference from RFC1738
but if it resolves Ç to the U+00C7 character, then it would indeed be a malformed URI, since that character needs to be %encoded
well, no, it weasels a bit on tht
this is slightly more difficult to explain without being able to paste the character in question
but then shouldn't be saying that "�..." as the fragment id is malformed rather than saying that "Ç..." as the fragment id is malformed?
http://www.faqs.org/rfcs/rfc3986.html is the canonical URL spec now
and it just says 'future URL tyeps'
it says "error: <...> escaping malformed URI reference"
"Çelik" is a valid fragment identifier
if you resolve the entity to the actual unicode character C with little curly tail thingy, as far as I can tell the specs say that is invalid and should be encoded as %C7
if you wanted to use "Çelik" as a fragment identifier, you would have to write it as "&Ccedil;elik"
well, that is the question - that get to the browser's internal representtion, whcih is not instntiated until someone clicks on it and sends it to the host
to validate HTML, you have to resolve the entities. when you resolve the entity, you wind up with a URI that violates the RFCs. HTML 4.01 refers to the RFCs to define what a URI is.
it's not a significant issue, but it seems fairly clear to me that the document is thus not valid
RobertBachmann is Robert Bachmann <http://rbach.priv.at/> and lives in Austria (Timezone: 01:00)
robert, looking at your markup for the channel transcript
I liek it except that there is only an id per time, not per line
yes that might be a problem in some cases. Other options?
maybe a monotonically increasing ID?
thought about a monotonic id back in november when I started hacking on LogBot, but timestamps looked nicer to me.
s/monotonic id/monotonically increasing ID/
maybe I could use "HHMMSS.SSS" instead of "HHMMSS". Of course I would need some naughty hack to ensure that ID's are unique: if (last_id == current_id) current_id += 0.001;
trovster is a web developer from the UK who writes on http://www.trovster.com and runs www.csslounge.co.uk
well, I'm thinking about a use case where this is dynmically updated by AHAH
so you'd effectively be passing an 'ack up to' back to the server
re the timestamps time of receipt? I forget in IRC
RobertBachman: what would be lost if there were no unique id's on the chat events?
ability to reference, and the use case I mention
[[to-do]] http://microformats.org/wiki?title=to-do&diff=0&oldid=4681 * DrErnie * (+960) Wiki-Thon Proposal
[[to-do]] http://microformats.org/wiki?title=to-do&diff=0&oldid=4682 * DrErnie * (+0) Agenda (Wishlist) -
Anyone want to move to Texas and work with me?
i'm already in texas, and sort of want to move away
:O
Are you in Austin?
[[to-do]] http://microformats.org/wiki?title=to-do&diff=0&oldid=4683 * DrErnie * (+181) Wiki-Thon Proposal -
kevinmarks: well, hmm, querying for time ranges works too, right? as long as the edges are well defined, bad case would be duplicate stamps (which is legit) where you get the first of the duplicates on one query, then the duplicate time stamp comes in, then you do the next query and miss it. some sort of count param?
the case i'm thinking of is a dynamicly updated page
(I have doen soem of this with greg elin before)
unique id's in a mircoformat seem particularly evil, all sorts of authorability and scoping issues
Indeed.
what you want is the client to send an async HTTP request saying in effect 'I have everything up to id 1234'
kevinmarks: or.. i have everything up to timestamp 00:00:03, with 2 entries at that last time stamp
and the server can block until it sees somethign new, or return everythign from 1234 up to current
and the server can block until it sees something new, or return everything from 00:00:03 onwards (except the first 2)
this is why I'm asking who makes the timestamps
if the server does using arrival time, that works
if the senders timestamp and out-of order arrival is repaired in the client, it is different
nd I don't remeber how IRC works in that respect
looks like mflogbot does it on the server
but i'm not sure that it matters
cks: ">RobertBachman: what would be lost if there were no unique id's on the chat events?" ID's must be uniqe in (X)HTML
hmm, that is, what if there were no id's at all? i see that there's an issue with update protocols, anything else?
so timestamps aren't in the protocol, therefore clients are manufacturing them
there are many possible chat protocols
regarding timestamps: for mflogbot the date of the server is used. (Date now = new Date();)
cks: If added the ID's because I wanted users to be able to link to a specific part of the chat log.
it means if you combine logs you have to regenerate all the id's, which is sort of hostile
also pretty much impossible to author by hand (although that might be less of an issue here?)
well, roberts IDs are derived from timestamps, so by hand isn't hard
though automation is esier
not if you're combining multiple streams that might have similiar timestamps
but... i suspect there are some good ways around the issues without forcing unique id's
well, roberts model works there, in that all with the same second shwo up at once
when I did syncing irc to mp3's that was how I did it too.
realistically, a 1-second lag is OK
is the stuff you did synching irc to mp3's online? chrism said something about the infamous backchannel incident...
[[events]] http://microformats.org/wiki?title=events&diff=0&oldid=4684 * RyanKing * (+72) added netsquared
[[events]] M http://microformats.org/wiki?title=events&diff=0&oldid=4685 * RyanKing * (+1) fixing typo
ok, so two reasons for unique ids: (1) makes some kinds of procotols easier to write and (2) people might want to refer to certain bits of the conversation via fragment identifiers
a couple of queries on the backend would do it for the protocol, even with retro-timestamps for some theoretical chat systems:
(a) give me this particular range, whatever you got at this point in time
I don't think this should be a part of the format.
There is way to big of a chance for collisions.
(b) give me some point in the past forward to the present, plus anything that came in "late" from the point backwards
yeah, and you don't need to for (1), i think
(2) is harder, because it really is a human-related thing
and a practical one, in that frag-id's are really the only way the browsers support to get inside the docs
cks: http://epeus.blogspot.com/2004_10_01_epeus_archive.html#109776385870386881
well, if you can have out-of-order arrival, time range doesn't work
right, so you have a second query with slightly different semantics
and then it does
Atamido: I'm not saying it should be a MUST, but it is something useful, and worth formulating good practice at least
it's not needed for the protocol stuff, but the human-related "i want to link to that comment" thing is harder
well, it makes the protocol stuff easier to do in a DRY way, but there are otehr ways, yes
unique id's suck unless there's some expectation that they're really identifying something unique in the universe, which doesn't seem to be the case for anything except maybe timestamp + nick + channel + server + hash-of-entire-comment
but circling around to the "want to reference bits of the conversation via a url", does that actually happen right now in practice?
hmm, the ILRT Bot RDF format spits out RDF:ID's
which are just the timestamps
but rdf has well defined semantics for merging set of files with file-unique id's
what kind of conversation are you talking about?
ah. IRC
the id's in :
pink numbers were better
<foaf:chatEvent rdf:ID="T00-14-11">
are in fact URIs
e.g. http://chatlogs.planetrdf.com/swig/2006-02-02.rdf#T00-14-11
using the implicit base uri of the doc is kind of a hack (there's not a base uri specified explicitly in the example on the chat-formats page)
no it isn't in the microformt context
URLs are globlly unique
so that whole issue is solved, and an id is defiend to be locally unique, so you re using both in their most natural way
using the implicit base uri in rdf is problematic, because if you move the document it changes, probably there should be an xml:base in there
well, thats because RDF likes URIs not URLs
[[hcard-brainstorming]] M http://microformats.org/wiki?title=hcard-brainstorming&diff=0&oldid=4686 * Chris Messina * (+87) Implied "FN and N" Optimization (proposal) -
[[hcard-brainstorming]] M http://microformats.org/wiki?title=hcard-brainstorming&diff=0&oldid=4687 * Chris Messina * (+13) Implied "FN and N" Optimization (proposal) -
anyway, the other thing it means is that you need to use a toolchain to merge documents, you can't just paste them together. that's ok for rdf, as it's expected, maybe not so good for microformats
I doubt there is a good way to create meaningful ID's that are surely unique, even within a single document..
Remember, ID's can't begin with a number.
And any writer becomes more complex if you have to ensure that two timecodes aren't the same.
(if you are using timecodes for the ID)
but, ok, that opens a question: if you'd like to point to part of a chat from the outside, what's a good way?
id in the doc, plus url#frag, is pretty slick...
Perhaps creating random 64bit IDs?
If you just need something to point at I mean.
"message-5f68a2e8
rnadom is stupid
incrementing with each message is fine
timestamps have wrapping issues
datestamps would be appropriate then (can ignore wrapping issues for the purposes of the diiscussion :-) )
that is, they wrap at the end of the universe...
These logs were automatically created by mflogbot on chat.freenode.net using a modified version of the Java IRC LogBot.
See http://microformats.org/wiki/mflogbot for more information.