10:53
<MikeSmith>
zcorpan: thanks for the latest set of xml-stylesheet test cases
10:54
<MikeSmith>
I think I've now got handling for all those behaving as expected
10:54
<MikeSmith>
when you have time, please test and let me know
10:54
<MikeSmith>
i'm going through the previous test cases now to make sure the latest changes didn't introduce any regressions
10:55
<MikeSmith>
oops
10:55
<MikeSmith>
get some build errors
10:59
<zcorpan>
MikeSmith: should the ...all/ checker enable the xml-stylesheet checker?
10:59
<zcorpan>
MikeSmith: iirc i had to type in the checker url manually when i tested
11:00
<MikeSmith>
zcorpan: I think the all/ checker enables it
11:01
<MikeSmith>
I'm not setting anything manually when I test
11:01
<MikeSmith>
I just set the preset to XHTML5
11:01
<zcorpan>
ok
11:02
<zcorpan>
i can't connect to qa-dev atm
11:05
<MikeSmith>
zcorpan: yep, I'm working on it .. had a build problem
11:05
<MikeSmith>
I should have it back in a 5 minutes or so
11:11
<MikeSmith>
zcorpan: OK, it's back
11:19
<MikeSmith>
zcorpan: I have checking for the charset pseudo-attribute in my workspace
11:19
<MikeSmith>
lemme know if you want me to push that up to qa-dev too
11:30
<zcorpan>
MikeSmith: <?xml-stylesheet href=?> gives Error: Duplicate xml-stylesheet pseudo-attribute href.
11:31
<MikeSmith>
hmm, OK
11:31
<MikeSmith>
will try to fix that now
11:34
<zcorpan>
MikeSmith: is it possible to have a look at the code?
11:35
<MikeSmith>
yeah, I can post a patch to bugzilla
11:35
<MikeSmith>
hang on a minute
11:37
<MikeSmith>
clearly I've somehow got it adding the attribute name twice for that case
11:40
<zcorpan>
MikeSmith: regarding multiple xslt pis, i think firefox ignores all but the *last* (but other browsers use the first)
11:40
<MikeSmith>
zcorpan: http://bugzilla.validator.nu/attachment.cgi?id=135
11:40
<zcorpan>
MikeSmith: thanks
11:40
<MikeSmith>
zcorpan: hmm, that's kind of bad
11:41
<zcorpan>
what's the windows1252 stuff?
11:43
<zcorpan>
<?xml-stylesheet href="" title="&#x0D;"?> should be valid
11:43
<zcorpan>
see XML LegalChar
11:44
<zcorpan>
(maybe xml 1.0 4ed LegalChar; ask hsivonen for advice)
11:44
<zcorpan>
if they're different
11:45
<zcorpan>
you might want to test "EOF" inside entity and ncr, too
11:45
<MikeSmith>
OK
11:45
<MikeSmith>
that stuff is all just from the htmlparser Tokenizer code
11:46
<MikeSmith>
so I probably need to tweak it further to align with XML spec
11:47
<MikeSmith>
man, I can see already I have some kruft around the adding of attributes that I need to clean up
11:50
<zcorpan>
<?xml-stylesheet href="" title="&"?> should be invalid
11:50
<MikeSmith>
hmm, yeah
11:50
<MikeSmith>
that one will be harder to fix
11:51
<MikeSmith>
I've got the duplicate-attribute thing fixed in my workspace
11:52
<MikeSmith>
just needed to remove conditionals around lines 482 and 518
11:53
<MikeSmith>
you want to me to push the change to qa-dev now?
11:53
<zcorpan>
naw it's ok
11:53
<zcorpan>
not sure it's good to single out checking for "<"
11:57
<zcorpan>
but maybe it is
11:59
<zcorpan>
yeah, it is, i'm just not thinking clearly
11:59
<MikeSmith>
OK
12:03
<zcorpan>
MikeSmith: type="TEXT/XML" ... should be case-insensitive
12:03
<MikeSmith>
OK
12:03
<MikeSmith>
I thought I had it that way already
12:03
<MikeSmith>
but I guess not
12:04
<zcorpan>
MikeSmith: and should maybe even recognize things like "text/xml; charset=utf-8"
12:04
<MikeSmith>
I think it does do that already, at least
12:04
<zcorpan>
but maybe that's hard and not worth the bother
12:04
<zcorpan>
oh
12:04
<MikeSmith>
I'm just using the existing HTML5 datatype library for checking that
12:05
<MikeSmith>
so if it doesn't work for the PI, it means it's broken elsewhere too
12:08
<zcorpan>
if ("application/xml".equals(attrValue)
12:09
<zcorpan>
this is for the multiple xslt whining
12:09
<MikeSmith>
yeah
12:09
<MikeSmith>
should I remove that?
12:10
<MikeSmith>
application/xml is the actual recommended mime type, isn't it?
12:10
<MikeSmith>
I do realize it doesn't work across browsers
12:10
<MikeSmith>
I will add the warning stuff after this
12:10
<MikeSmith>
with a link to the wiki page
12:10
<zcorpan>
my point is ideally it should be case-insensitive and support parameters
12:11
<zcorpan>
so <?xml-stylesheet href="" type="TEXT/XML"?><?xml-stylesheet href="" type="application/xml; charset=utf-8"?> should give a message about multiple xslt pi
12:11
<MikeSmith>
ah, OK
12:15
<MikeSmith>
so I could lowercase the string before comparing and then just use startsWith()
12:15
<MikeSmith>
I think
12:19
<zcorpan>
type="text/xslLOL"
12:19
<zcorpan>
just lowecasing is probably good enough
12:20
<zcorpan>
gotta go
23:45
<aho>
webgl behaves like flash with wmode=transparent. this is pretty slow and usually totally unnecessary. are there any plans for some kind of switch which would make it behave more like an opaque overlay?