XML Not Working in Explorer 6

Rails
1

I’m playing with XML using Builder but seem to be having some trouble in
Explorer 6. My sample code:

xml = Builder::XmlMarkup.new
xml.instruct!
xml.instruct! ‘xml-stylesheet’, :href=>‘/stylesheets/style-xml.css’,
:type=>‘text/css’
xml.declare! ‘DOCTYPE’, :html,
:PUBLIC,
“-//W3C//DTD XHTML 1.0 Transitional//EN”,
“http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd”
xml.html ( :xmlns=>“XHTML namespace”, :lang=>“en-US”,
‘xml:lang’=>“en-US” ) {
xml.head {
xml.title ‘My test page’
}
xml.body {
xml.comment ‘Test comment’
xml.br
xml.p
xml.em ‘TEST!’
}}

This displays as expected under FireFox. But under Explorer 6 the line
breaks don’t display and the title shows up as regular text. I copied
the document headers from an old Perl/CGI app that works fine under
Explorer. The stylesheet is read and contains definitions for body,
comment and em only.

Can anyone point out where I’ve gone wrong?

Kirk

2

On Tuesday 27 Dec 2005 22:01, Kirk Bocek wrote:

I’m playing with XML using Builder but seem to be having some trouble in
Explorer 6. My sample code:
[SNIP]
This displays as expected under FireFox. But under Explorer 6 the line
breaks don’t display and the title shows up as regular text. I copied
the document headers from an old Perl/CGI app that works fine under
Explorer. The stylesheet is read and contains definitions for body,
comment and em only.

I would guess it’s down to IE only supporting XHTML when it’s served as
text/html, and that by using a .rxml template, you’ll be serving it as
application/xml.

See this:

For XHTML documents in Rails, you can (or most probably “should”) use
regular .rhtml files, which are then served as text/html - this works
fine
for me using XHTML 1.0 Strict.

Or I guess you could use .rxml templates and munge the headers in your
controller, using:

headers[‘Content-Type’] = “text/html”

I haven’t actually tried that myself, although did experiment sending
some
other .rxml stuff I was generating as text/plain in order to read it
straight
in the browser, which worked, so I can think of no reason why the above
suggestion wouldn’t.

None of these are particularly ideal solutions - the best one would be
to
eliminate Internet Explorer from existence, or at least until such time
as it
becomes a real browser and keeps well up-to-date with the fast-moving
standards! :wink:

For XHTML 1.0 both transitional and strict, the specs allow you to serve
it as
text/html, even though application/xhtml+xml would be more correct.
That is
only mandatory with XHTML 1.1, which is why IE can’t handle XHTML 1.1
when
served properly.

This may not be your exact problem, but I’m sure it’s a contributing
factor.
In the very least, you shouldn’t be serving XHTML as just
application/xml,
because that’s a bit on the vague side.

IE can handle XML a bit (AFAIK, but I rarely use IE these days), but I
don’t
know what it would do if you sent it XHTML as application/xml - my guess
is
it would choke rather horribly, in some way similar to what you are
seeing at
the moment. :wink:

Hope this helps!

~Dave

–

Dave S.
Rent-A-Monkey Website Development
Web: http://www.rentamonkey.com/

3

Thanks for the reply, Dave.

Dave S. wrote:

I would guess it’s down to IE only supporting XHTML when it’s served as
text/html, and that by using a .rxml template, you’ll be serving it as
application/xml.

See this:

Internet Explorer - Wikipedia

Excellent link! I knew Explorer would have problems. This is a nice
summary of all of them.

For XHTML documents in Rails, you can (or most probably “should”) use
regular .rhtml files, which are then served as text/html - this works
fine for me using XHTML 1.0 Strict.

Yep. Works for me too. I’m playing with .rxml because it was suggested
as a preferred way to progmatically generate web pages as opposed to
using <% %>. I’ve done it forever using Perl’s CGI module. Trying to
figure out how to do it in Rails.

Or I guess you could use .rxml templates and munge the headers in your
controller, using:

headers[‘Content-Type’] = “text/html”

Well, this does seem to cause both browsers to act differently. Both
seem to ignore <?xml-stylesheet …> They seem to read <LINK …> but do
strange things. Maybe only embedded styles will work?

None of these are particularly ideal solutions - the best one would be
to
eliminate Internet Explorer from existence

Sure, but if we’re writing an app for public consumption…

IE can handle XML a bit

Yep, it was partially working before I made your Content-Type change. If
I can figure out how to set the page title and get paragraph breaks, it
would work fine.

Thanks for the suggestion, though, it’s gotten me one step further
along.

Kirk

4

Dave S. wrote:

by using a .rxml template, you’ll be serving it as
application/xml.

Oh, and it actually it comes over as ‘text/xml’

Kirk

5

Dave S. wrote:

So essentially, you need to use content negotiation to give the
browser what it prefers.

Nope. No way. Nuh-Uh.

Whatever, you’re not going to be able to get this to work in IE using
xml-stylesheet directives, AFAIK, because IE sucks at this kind of
stuff.

Gonna spend a little time on this. Explorer seems to download the
stylesheet correctly as ‘text/xml’ but not ‘text/html’.

… although then I guess you’d still need content negotiation

I said no and I mean it!

Ultimately, I think you can do what you want to do by sending it as
text/html,
and using regular XHTML embedded stylesheet stuff rather than
xml-stylesheet directives.

That’s my next test. At this point it looks like I’ve duplicated all the
headers from the Perl app that works fine in Explorer and uses embedded
styles. Stylesheets are the only difference.

Or figure out content negotiation in Rails,

Now am I going to have to get physical here?..

I reckon it would actually be quite
easy (and undoubtedly neater - PHP kinda grosses me out these days!)

Eeew! I hate it when you step in a pile of syntax!

Easiest method, I reckon, is just to send it as text/html from your
controllers, using .rxml views, and putting standard HTML embedded
stylesheet stuff in them.

That the way I was leaning, but I still want to figure out what I can
and cannot do in Explorer.

Kirk

6

On Wednesday 28 Dec 2005 19:31, Kirk Bocek wrote:

Well, this does seem to cause both browsers to act differently. Both
seem to ignore <?xml-stylesheet …> They seem to read <LINK …> but do
strange things. Maybe only embedded styles will work?

Yes, I believe so - it’s IE that’s the problem.

The trouble is, what works for Firefox and other modern browsers will
not work
for IE. So essentially, you need to use content negotiation to give the
browser what it prefers. I’ve done this in PHP a few times (actually,
my
main www.rentamonkey.com site does it) to serve XHTML 1.1 to browsers
that
will accept it properly, but change it to XHTML 1.0 as text/html for
those
that won’t. I’m not sure it’s an ideal solution, or even worth it, but
I
wanted to try it out somewhere so that was as good a place as any to do
it.

I haven’t figured out content negotiation in Rails yet (not that I’ve
tried),
but for now am happy enough with XHTML 1.0 Strict served as text/html.

Whatever, you’re not going to be able to get this to work in IE using
xml-stylesheet directives, AFAIK, because IE sucks at this kind of
stuff.

Sure, but if we’re writing an app for public consumption…

I know, it was a joke… didn’t you see the smiley? :wink: :wink:

In web developer utopia, there would be no Internet Expletive!

Yep, it was partially working before I made your Content-Type change. If
I can figure out how to set the page title and get paragraph breaks, it
would work fine.

No idea, but I guess you need to check out how IE handles pure XML
stuff,
rather than application/xhtml+xml, although then I guess you’d still
need
content negotiation, or you could send it all as text/xml (saw your
other
email, my bad, I didn’t check it before, just assumed .rxml sent as
application/xml) but that might have weird results.

Ultimately, I think you can do what you want to do by sending it as
text/html,
and using regular XHTML embedded stylesheet stuff rather than
xml-stylesheet
directives.

Or figure out content negotiation in Rails, then set the page up as
appropriate. Incidentally, if you’d like to see how I did it in PHP,
check
here: http://www.rafb.net/paste/results/cyGrz844.html

That’s the code from www.rentamonkey.com. I reckon it would actually be
quite
easy (and undoubtedly neater - PHP kinda grosses me out these days!) to
convert that to Rails, so feel free to if you like ALTHOUGH I’m still
not
sure there’s actually any point in doing that, other than for
I’m-bleeding-edge-nyah-nyah-ness. :wink:

Thanks for the suggestion, though, it’s gotten me one step further
along.

No worries, will be interested to see if you get it working!

Easiest method, I reckon, is just to send it as text/html from your
controllers, using .rxml views, and putting standard HTML embedded
stylesheet
stuff in them. Kinda gets you the best of all worlds, currently, in a
way
that will work with all browsers.

~Dave

–

Dave S.
Rent-A-Monkey Website Development
Web: http://www.rentamonkey.com/

7

On Wednesday 28 Dec 2005 20:35, Kirk Bocek wrote:

Nope. No way. Nuh-Uh.
I said no and I mean it!
Now am I going to have to get physical here?..

Are you sure you haven’t mistaken content negotiation for user-agent
sniffing?
Just in case you thought I was advocating user-agent sniffing (“pure
evil”),
they’re two rather different things.

As far as I know, there’s nothing inherently bad with content
negotiation. If
you’re trying to do something a bit more bleeding edge and it’s not
supported
by older browsers (and you have an easy way of downgrading it to give
them
something that works), then there’s nothing inherently bad about that,
IMO.

Browser sniffing on the other hand… eouw!

Not that the thing I did on www.rentamonkey.com is particularly good
code, or
even has any point to it at the moment - it was slightly experimental,
but I
wanted to do it with a view to being fully XHTML 1.1 for clients that
support
it, and eventually dropping XHTML 1.0 at some point, although I suspect
I’ll
have re-built that site in Rails long before that ever happens.

Gonna spend a little time on this. Explorer seems to download the
stylesheet correctly as ‘text/xml’ but not ‘text/html’.

I was going to ask why you’re sending the stylesheet as anything but
text/css
(which would work in IE, but not Gecko-based browsers), but then I
realised
you meant that the stylesheet would download when the page was sent as
text/xml but not text/html, and that you were sending the stylesheet as
text/css all along. :slight_smile:

That’s my next test. At this point it looks like I’ve duplicated all the
headers from the Perl app that works fine in Explorer and uses embedded
styles. Stylesheets are the only difference.

I don’t know if you’re using xml-stylesheet directives or standard HTML
embedded stylesheet tags, but I would suggest you probably should send
the
page as text/html and use regular stylesheet tags, not xml-stylesheet
directives.

That should definitely work, and is the same as what a .rhtml template
would
do, with the exception that you’re using .rxml to generate it, and the
controller to change the content type of the page.

I think the problem is that, AFAIK, IE won’t work with XML using a
text/css
stylesheet, because it’s expecting a text/xsl one, and I guess that’s
one of
the things you’re coming up against.

I could be wrong though - it’s been a while since I’ve done any of this
stuff,
and especially not in IE.

Eeew! I hate it when you step in a pile of syntax!

at the things I used to do in PHP.

That the way I was leaning, but I still want to figure out what I can
and cannot do in Explorer.

Sounds to me like you’ve figured it out already . :wink:

~Dave

–

Dave S.
Rent-A-Monkey Website Development
Web: http://www.rentamonkey.com/

8

A short update:

Explorer does download style sheets through the <link …> tag. It
does not, however, use the <?xml-stylesheet … > declaration.

This code in my .rxml document is working:

xml.head {
xml.title 'My XML Builder Test Page'
xml.link :type=>"text/css", :rel=>"stylesheet",

href=>"/stylesheets/style-xml.css"
}

Kirk

9

Dave S. wrote:

Are you sure you haven’t mistaken content negotiation for user-agent
sniffing?
Just in case you thought I was advocating user-agent sniffing (“pure
evil”),
they’re two rather different things.

Yes, I did make that mistake. I assumed they were two aspects of the
same solution.

Sounds to me like you’ve figured it out already . :wink:

You’re too kind. No really – be less kind to me.

Kirk