Options:
- # Session Start: Tue Oct 09 00:00:00 2007
- # Session Ident: #html-wg
- # [00:00] <Hixie> Dashiva: it's not a matter of failing cheaply for the web authors
- # [00:00] <Hixie> Dashiva: it's a matter of failing cheaply for the users and implementors
- # [00:02] <Dashiva> But whether it ends up failing is up to the author. And I don't see how authors are likely to fail just this function enough to make a difference, without also having generally poor performance code
- # [00:02] * Joins: mjs_ (mjs@66.43.120.234)
- # [00:02] * Quits: mjs (mjs@66.43.120.234) (Ping timeout)
- # [00:03] <Hixie> i don't understand what site is going to actually want to get all elements this way
- # [00:03] <Hixie> i understand your theoretical use case
- # [00:03] <Hixie> but i don't see any evidence that that would ever happen
- # [00:04] <anne> can't we introduce * if the need is there?
- # [00:04] <Hixie> we already have * in at least two other APIs
- # [00:06] * Quits: zcorpan_ (zcorpan@83.227.34.9) (Ping timeout)
- # [00:10] * Quits: tH (Rob@87.102.7.125) (Quit: ChatZilla 0.9.78.1-rdmsoft [XULRunner 1.8.0.9/2006120508])
- # [00:10] <Dashiva> If it's not '', I don't see a point in allowing '*' since you might as well use gEBTN when you're special-casing the empty set
- # [00:12] <Hixie> agreed
- # [00:12] <kingryan> Hixie: commit.pl modified and replied, hopefully it'll be helpful
- # [00:13] <Hixie> thanks
- # [00:13] * Joins: zcorpan_ (zcorpan@83.227.34.9)
- # [00:14] <kingryan> btw, my connection to http://www.whatwg.org/issues/listen keeps timing out. is there anyone besides hixie who can look at that?
- # [00:16] <anne> nope, Hixie handles that part of whatwg.org
- # [00:16] <Hixie> how do you mean, timing out?
- # [00:17] <anne> although it seems to work for me (trying with wget)
- # [00:17] <zcorpan_> <div class="*"> .. getElementsByClassName("*")
- # [00:17] <kingryan> it never finishes loading and safari reports it as a timeout
- # [00:17] <Hixie> zcorpan_: that'll work
- # [00:18] <kingryan> (safari 3)
- # [00:18] <Hixie> kingryan: it's not supposed to finish loading.
- # [00:18] <Hixie> kingryan: not sure what "timeout" safari3 is hitting though
- # [00:18] <zcorpan_> Hixie: yep
- # [00:20] <Hixie> kingryan: (fyi, when changing system {} '', '' to ``, you have to drop the first argument)
- # [00:20] <kingryan> ah yes
- # [00:20] <kingryan> I changed that in my test script, but forgot to backport it
- # [00:21] <Hixie> though i never really feel safe running `` code, i always think i'm gonna be passing something to a shell
- # [00:21] <jgraham> anne: Not today, went to see a film instead. "Control". Highly recommended. Now - sleep.
- # [00:22] <Hixie> also you forgot quotes around the argument :-)
- # [00:22] <kingryan> Hixie: I, also feel weird using it, but it's the only shell scripting interface in perl (that I know of) that lets you capture the STDOUT
- # [00:22] <Hixie> i wrote my own once
- # [00:22] <Hixie> to get around this
- # [00:22] <Hixie> called safeBackticks or something
- # [00:22] <Hixie> i can't use this though
- # [00:22] <kingryan> why not?
- # [00:22] <Hixie> because it would break if the message contained quotes
- # [00:23] <kingryan> ah yeah
- # [00:23] * Hixie adds some code to escape them
- # [00:23] <Hixie> i have no idea what kind of escaping i need here
- # [00:24] <Hixie> i don't know what shell is being used
- # [00:24] * Hixie passes the commit message by file instead
- # [00:24] <Philip> open $fh, '-|', 'program', 'arg1', arg2'; ?
- # [00:24] <Hixie> Philip: does that run 'program' via a shell, or directly, as in system {} ?
- # [00:25] <kingryan> http://perldoc.perl.org/perlop.html#Regexp-Quote-Like-Operators sugggests using the Text::Balanced module
- # [00:25] * Quits: mjs_ (mjs@66.43.120.234) (Ping timeout)
- # [00:28] <Hixie> perlipc helpfully says:
- # [00:28] <Hixie> Another common use for this construct is when you need to execute some-
- # [00:28] <Hixie> thing without the shell's interference. With system(), it's straight-
- # [00:28] <Hixie> forward, but you can't use a pipe open or backticks safely. That's
- # [00:28] <Hixie> because there's no way to stop the shell from getting its hands on your
- # [00:28] <Hixie> arguments. Instead, use lower-level control to call exec() directly.
- # [00:28] <Hixie> (and gives some gnarly examples of what they mean)
- # [00:29] <Hixie> (which is what i based that aforementioned implementation on)
- # [00:29] <Philip> Hixie: I can't find anything obvious in the documentation to say what happens with list-open, but it seems to work in the practice at avoiding the shell
- # [00:29] <Hixie> Philip: i just found a paragraph in perlipc that says that it's the way to avoid a shell invokation, and that's it's new as of 5.8.0
- # [00:29] <Philip> e.g. "open $fh, '-|', 'echo', '$PATH'" says "$PATH", whereas "open $fh, '-|', 'echo $PATH'" prints the actual path
- # [00:29] <Hixie> which postdates my previous attempt at this
- # [00:30] <Philip> Aha, okay
- # [00:31] <Hixie> well, that was far more complicated than necessary to do what kingryan was trying to do
- # [00:33] <Hixie> ok well we'll see what happens now
- # [00:43] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
- # [01:05] * Quits: zcorpan_ (zcorpan@83.227.34.9) (Ping timeout)
- # [01:09] * Joins: sbuluf (shaf@200.49.140.233)
- # [01:30] * Quits: hasather (hasather@90.227.221.48) (Quit: leaving)
- # [01:40] * Joins: olivier (ot@128.30.52.30)
- # [02:29] * Joins: karl (karlcow@128.30.52.30)
- # [02:47] * Quits: aroben (adamroben@17.203.15.247) (Quit: aroben)
- # [03:34] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [03:43] * Quits: kingryan (rking3@208.66.64.47) (Quit: kingryan)
- # [04:37] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
- # [04:45] <karl> http://www.caroline-middlebrook.com/blog/warning-dont-overlook-wordpress-blog-validation
- # [04:45] <karl> "The Image Description field populates the ALT tag as I have shown. This is a nice place to put keywords as long as you don’t overdo it."
- # [04:45] <karl> ????
- # [05:56] * Joins: Zeros (Zeros-Elip@69.140.40.140)
- # [06:09] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [06:29] * Joins: dbaron (dbaron@71.204.145.103)
- # [07:00] <karl> anne: do you know this? http://www.w3.org/2001/01/qa-ws/pp/alex-rousskov-measfact.html
- # [07:01] <karl> and this http://httpd.apache.org/test/
- # [07:17] * Quits: marcos (chatzilla@131.181.148.226) (Ping timeout)
- # [07:18] * Joins: marcos (chatzilla@131.181.148.226)
- # [07:31] * Joins: aroben (adamroben@67.160.250.192)
- # [07:49] * Quits: aroben (adamroben@67.160.250.192) (Client exited)
- # [07:49] * Joins: aroben (adamroben@67.160.250.192)
- # [08:03] * Quits: Zeros (Zeros-Elip@69.140.40.140) (Quit: Leaving)
- # [10:08] * Disconnected
- # [10:08] * Attempting to rejoin channel #html-wg
- # [10:08] * Rejoined channel #html-wg
- # [10:08] * Topic is 'HTML WG http://www.w3.org/html/wg/ also logged: http://krijnhoetmer.nl/irc-logs/'
- # [10:08] * Set by DanC on Thu Sep 06 23:39:27
- # [10:11] * Parts: Shunsuke (Shunsuke@123.176.107.50)
- # [10:32] * Quits: Lionhear1 (robin@66.57.69.65) (Connection reset by peer)
- # [10:33] * Joins: Lionheart (robin@66.57.69.65)
- # [10:35] * Joins: paullewis_ (paullewis@81.255.115.137)
- # [10:35] * Quits: paullewis (paullewis@81.255.115.137) (Connection reset by peer)
- # [10:45] * Joins: hasather (hasather@90.227.221.48)
- # [10:45] * Joins: zcorpan_ (zcorpan@88.131.66.80)
- # [10:53] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
- # [10:54] * Quits: jmb (jmb@152.78.68.189) (Ping timeout)
- # [10:54] * Joins: jmb (jmb@152.78.68.189)
- # [10:55] * Joins: ROBOd (robod@89.122.216.38)
- # [11:11] * Quits: Philip (philip@80.177.163.133) (Ping timeout)
- # [11:13] * Quits: Lachy (Lachy@213.236.208.22) (Connection reset by peer)
- # [11:13] * Joins: Lachy (Lachy@213.236.208.22)
- # [11:13] * Quits: Lachy (Lachy@213.236.208.22) (Connection reset by peer)
- # [11:13] * Joins: Lachy (Lachy@213.236.208.22)
- # [11:15] * Quits: Lachy (Lachy@213.236.208.22) (Quit: Leaving)
- # [11:15] * Joins: Shunsuke (Shunsuke@123.176.107.50)
- # [11:37] * Joins: Lachy (Lachy@213.236.208.22)
- # [11:44] * Quits: jmb (jmb@152.78.68.189) (Ping timeout)
- # [11:47] * Joins: jmb (jmb@152.78.68.189)
- # [12:00] * Quits: hasather (hasather@90.227.221.48) (Quit: leaving)
- # [12:02] * Joins: Philip (philip@80.177.163.133)
- # [12:07] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
- # [12:49] * anne wonders if Yves is suggesting that IPV and TCP have the same interoperability issues as the rest of the Web standards
- # [12:49] * anne wouldn't be surprised
- # [12:50] * Quits: sbuluf (shaf@200.49.140.233) (Ping timeout)
- # [12:52] <anne> DanC, you might want to update http://www.w3.org/2007/app to point to RFC 5023
- # [13:22] * hsivonen takes the ostrich approach to data templates for now
- # [13:37] <anne> hsivonen, what's the ostrich approach and I'm not sure I understand the concern about specialcasing GET access requests
- # [13:38] <anne> the reason there's a GET access request is that for XML resources this gives the author the ability to deny certain domains without server support
- # [13:38] <hsivonen> anne: the ostrich approach is putting my head in the sand and pretending the issue does not exist
- # [13:38] <anne> the other reason is that for the author supporting GET is the most trivial
- # [13:38] <anne> hsivonen, ah, duh :)
- # [13:39] <hsivonen> anne: making GET responses stick in the cache when RFC 2616 says otherwise would seriously suck in the following case:
- # [13:39] <hsivonen> 1) script GETs /foo
- # [13:39] <hsivonen> 2) script wants to POST to /foo
- # [13:40] <hsivonen> 3) XHR GETs /foo with method check and pins the response to cache
- # [13:40] <hsivonen> 4) XHR performs the POST
- # [13:40] <hsivonen> 5) Script GETs the resource and gets a stale copy
- # [13:40] <hsivonen> script dev curses, appends salt and all cache benefits are lost
- # [13:41] <anne> I think we should be able to specify it in a way that for 5 you get a fresh copy
- # [13:41] <anne> but for subsequent POST requests the cached copy is used
- # [13:41] <hsivonen> anne: now, if you special-case the GET response cache pinning so that any non-access-control GET invalidates the cache lazily,
- # [13:42] <hsivonen> anne: wouldn't you be better off specifying this on a layer above HTTP
- # [13:42] <hsivonen> anne: instead of surgically altering HTTP caching behavior
- # [13:42] <hsivonen> anne: i.e. leaking your spec to the HTTP library
- # [13:43] * anne is trying not to feel responsible as he didn't come up with the solution
- # [13:43] <hsivonen> anne: how's your suggesting better than maintaining a separate access-control cache that wouldn't even require reparsing of the cached GET response?
- # [13:43] <anne> I suppose the layer above HTTP is OPTIONS?
- # [13:43] <hsivonen> no, OPTIONS is on HTTP layer
- # [13:44] <anne> right, that's what I thought, so what's your solution?
- # [13:44] <hsivonen> but remembering that /foo allows XHR POST is above HTTP
- # [13:44] <hsivonen> my solution is maintaining a method permission table that is independent of the HTTP cache
- # [13:45] <anne> my solution was maintaining an independent cache for access requests
- # [13:45] <anne> to not lose benefits of the HTTP cache
- # [13:46] <hsivonen> anne: do you mean a separate HTTP caching context that within itself behaves like a normal HTTP cache?
- # [13:46] <anne> I think so, so it honers HTTP headers related to caching
- # [13:47] <hsivonen> specifying that access-control checks happen in a separate HTTP caching context would solve the problem if browser devs are willing to built that capability
- # [13:48] <hsivonen> do browsers already have side-by-side independent HTTP caches?
- # [13:48] <anne> we do for widgets at least
- # [13:49] <anne> a simpler solution than that might be good as well though
- # [13:49] <anne> i'm just not sure what to say about when that "access allowed" thingie would expire, etc.
- # [13:50] <hsivonen> yes, that's an issue.
- # [13:50] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
- # [13:50] <hsivonen> mainly for testing apps, though
- # [13:51] <hsivonen> once you get the permissions right, expiry until at the end of browser process might works
- # [13:52] <anne> hmm
- # [13:52] <hsivonen> or you could have a dedicated time-to-live header to go with the permission
- # [13:56] <zcorpan_> "in the upcoming release of the Markup Validator, detected usage of shorthand markup will be signaled as a warning." -- http://www.w3.org/QA/2007/10/shorttags.html
- # [13:59] * hsivonen mumbles something about GETsful software that can only do GET and POST
- # [13:59] <anne> hsivonen, I'm posting something to public-appformats now with the four solutions
- # [13:59] <hsivonen> anne: ok
- # [14:01] <hsivonen> hmm. is <base> now intentionally allowed in XHTML5?
- # [14:01] <anne> I should probably introduce the term "access request check" or something
- # [14:01] <hsivonen> yes
- # [14:03] <hsivonen> so is <base href=''/> considered to apply between the HTTP base URI and the root element xml:base in the inheritance chain?
- # [14:04] <anne> <base> is the Document.baseURI
- # [14:04] <anne> so yes
- # [14:04] <hsivonen> fun
- # [14:04] <zcorpan_> hsivonen: yes (re xhtml5)
- # [14:04] <hsivonen> Hixie: "A base element, if it has an href attribute, must come before any other elements in the tree that have attributes with URIs." seems tautologous to me
- # [14:04] <hsivonen> Hixie: is it?
- # [14:05] <hsivonen> Hixie: that is, if the only start tags that can conformingly occur before are <html>, <head> and <meta charset>
- # [14:05] <hsivonen> and profile is no more and never was a normal URI
- # [14:06] <zcorpan_> (or perhaps the intent was to relax the requirement so that you could e.g. place <title> before the <base>)
- # [14:07] <hsivonen> time to send mail, I guess
- # [14:09] * Parts: anne (annevk@81.68.67.12)
- # [14:10] * Joins: anne (annevk@81.68.67.12)
- # [14:36] <anne> hsivonen, if we pick option 4 (re: public-appformats) it would maybe be good to also invalidate the stored option if the access check of the POST request failed
- # [14:38] <hsivonen> yeah
- # [14:38] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [14:45] * hsivonen notes weird rel and relList duplication in the DOM
- # [14:46] <anne> duplication?
- # [14:52] <hsivonen> anne: the spec says that the rel and relList DOM properties must reflect the rel content attribute
- # [14:55] <anne> that seems about right
- # [14:56] <anne> well, except for the word properties
- # [14:56] <hsivonen> word properties?
- # [14:56] <anne> "properties" -> "attributes"
- # [14:57] <hsivonen> oh. right
- # [14:58] <anne> note that reflect is defined based on the type of attribute
- # [14:58] <anne> so for relList it means something different
- # [14:58] * Quits: Lachy (Lachy@213.236.208.22) (Quit: Leaving)
- # [14:59] <anne> (type of DOM attribute)
- # [15:00] <hsivonen> oh
- # [15:03] <zcorpan_> hsivonen: re <meta>, i think the intent is to allow <meta name> in other languages where metadata elements are expected
- # [15:04] <hsivonen> zcorpan_: that intent isn't at all clear
- # [15:04] <hsivonen> to me at least
- # [15:16] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
- # [15:30] * Joins: matt (matt@128.30.52.30)
- # [15:40] * Joins: kazuhito (kazuhito@222.151.148.141)
- # [15:53] * Joins: myakura (myakura@221.191.107.83)
- # [16:00] * Joins: Lachy (Lachy@213.236.208.22)
- # [16:36] * Joins: billmason (billmason@69.30.57.156)
- # [16:47] * Joins: tH_ (Rob@87.102.7.125)
- # [16:47] * tH_ is now known as tH
- # [16:58] * Parts: Lionheart (robin@66.57.69.65)
- # [17:00] <zcorpan_> wonder if i should test all aspects of [[Get]] on HTMLDocument and Window, and how they interact with functions, methods, properties and attributes
- # [17:01] <zcorpan_> e.g. <img name=getElementsByTagName>
- # [17:09] * Joins: aroben (adamroben@67.160.250.192)
- # [17:21] * Quits: Bob_le_Pointu (blp@80.248.208.232) (Ping timeout)
- # [17:25] * Quits: aroben (adamroben@67.160.250.192) (Quit: aroben)
- # [17:25] * Joins: Bob_le_Pointu (blp@80.248.208.232)
- # [17:28] * Quits: kazuhito (kazuhito@222.151.148.141) (Quit: Quitting!)
- # [17:31] * Joins: aaronlev (chatzilla@66.31.86.217)
- # [17:45] <anne> Are there any outstanding comments on <video> not having height=/width=?
- # [17:45] <anne> hsivonen, you might have raised that already?
- # [17:53] <zcorpan_> anne: doesn't seem like it
- # [17:54] <zcorpan_> http://canvex.lazyilluminati.com/misc/cgi/issues.cgi/folder/graphics-video
- # [17:54] * Quits: myakura (myakura@221.191.107.83) (Quit: Leaving...)
- # [17:54] * zcorpan_ searched through the "View all messages from this folder." page
- # [17:57] * Philip wonders if Google has indexed all the messages there
- # [17:58] <zcorpan_> http://www.google.com/search?q=site:canvex.lazyilluminati.com/misc/+cgi/issues.cgi&filter=0
- # [18:00] <Philip> Hmm, I guess that's a no
- # [18:02] <zcorpan_> http://canvex.lazyilluminati.com/misc/cgi/issues.cgi/message/%3C19700101011405970191.5386b4d6%40empyree.org%3E
- # [18:03] <Philip> Yahoo is a bit better, with about forty pages (out the two(?) thousand in total)
- # [18:03] <Philip> zcorpan_: Blame Hixie for that :-)
- # [18:03] <zcorpan_> :)
- # [18:10] <zcorpan_> Philip: could you have different <title>s for the pages?
- # [18:11] <zcorpan_> the subject line as <title>
- # [18:12] <zcorpan_> or the folder name for folders :)
- # [18:12] * Quits: paullewis_ (paullewis@81.255.115.137) (Quit: paullewis_)
- # [18:13] * Joins: dbaron (dbaron@63.245.220.241)
- # [18:18] * zcorpan_ doesn't find his color attribute parsing feedback
- # [18:19] <Philip> I guess it'd be nice to show the subject rather than ID on the message pages, but I can't see a trivial way to get that information (since '/issues/getdata body $id' only returns the body)
- # [18:20] * Quits: aaronlev (chatzilla@66.31.86.217) (Connection reset by peer)
- # [18:20] <Philip> Hixie: Would it be possible to make the issues API return the subject line for a given message ID?
- # [18:22] <Philip> zcorpan_: It does <title>s now, though with the message ID on the message pages
- # [18:24] <zcorpan_> ok
- # [18:26] * Joins: polin8 (polin8@75.71.72.175)
- # [18:32] * Joins: Sander (svl@86.87.68.167)
- # [18:51] * Joins: icaaq (icaaaq@217.13.228.226)
- # [18:58] * Quits: icaaq (icaaaq@217.13.228.226) (Ping timeout)
- # [19:02] * Joins: mjs (mjs@64.81.48.145)
- # [19:15] * Joins: paullewis (paullewis@82.242.109.217)
- # [19:18] <hsivonen> anne: I don't remember if I've raised video height/width in email, but it have discussed it on IRC
- # [19:20] <hsivonen> anne: I don't find a record of me raising it on-list
- # [19:20] * Joins: kingryan (rking3@64.81.240.149)
- # [19:48] * Joins: kevinw (kevinw@67.9.74.115)
- # [19:53] * Joins: aroben (adamroben@17.203.15.247)
- # [20:38] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
- # [20:45] * Joins: mjs (mjs@64.81.48.145)
- # [21:00] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
- # [21:00] * Quits: zcorpan_ (zcorpan@88.131.66.80) (Ping timeout)
- # [21:28] <Hixie> Philip: yeah, that should be possible... what level API? the server, or getdata?
- # [21:30] <Philip> Hixie: getdata (since I think that was the only documented API when I wrote this - I didn't even know there was another API until I looked just now)
- # [21:30] * Quits: kingryan (rking3@64.81.240.149) (Connection reset by peer)
- # [21:35] <Hixie> Philip: ok, added
- # [21:43] <Philip> Hixie: Thanks!
- # [21:43] <Philip> zcorpan: It shows the subject in the <title>/<h1> now
- # [22:05] * Joins: hober (ted@68.107.112.172)
- # [22:25] * Joins: mjs (mjs@17.255.98.11)
- # [22:25] * Quits: ROBOd (robod@89.122.216.38) (Quit: http://www.robodesign.ro )
- # [22:27] * Joins: mjs_ (mjs@17.203.14.209)
- # [22:28] * Quits: mjs (mjs@17.255.98.11) (Ping timeout)
- # [22:46] * Quits: matt (matt@128.30.52.30) (Quit: matt)
- # [23:41] * Joins: Lionheart (robin@66.57.69.65)
- # [23:43] * Quits: hober (ted@68.107.112.172) (Quit: ERC Version 5.3 (devel) (IRC client for Emacs))
- # [23:53] * Quits: laplink (link@193.157.66.199) (Ping timeout)
- # [23:53] * Quits: xover (xover@193.157.66.5) (Ping timeout)
- # [23:55] * Joins: xover (xover@193.157.66.5)
- # [23:55] * Joins: laplink (link@193.157.66.199)
- # Session Close: Wed Oct 10 00:00:00 2007
The end :)