Bug in the pre-fetching test

30 views
Skip to first unread message

Jonathan Klein

unread,
Jul 14, 2014, 11:06:17 AM7/14/14
to browse...@googlegroups.com
Hello,

Browserscope reports that IE11 doesn't support link rel="prefetch", but it does, just not in subframes.  See the discussion here: https://twitter.com/ericlaw/status/488700258373758976.  Is it possible to modify the test to expose this distinction?  

Cheers,

Jonathan

Lindsey Simon

unread,
Jul 15, 2014, 4:52:33 PM7/15/14
to browse...@googlegroups.com
On Mon, Jul 14, 2014 at 8:06 AM, Jonathan Klein <jonathan...@gmail.com> wrote:
Hello,

Browserscope reports that IE11 doesn't support link rel="prefetch", but it does, just not in subframes.  See the discussion here: https://twitter.com/ericlaw/status/488700258373758976.  Is it possible to modify the test to expose this distinction?  

Actually, I don't believe so since the test harness is done with frames. =(
 

Cheers,

Jonathan

--
You received this message because you are subscribed to the Google Groups "Browserscope" group.
To unsubscribe from this group and stop receiving emails from it, send an email to browserscope...@googlegroups.com.
To post to this group, send email to browse...@googlegroups.com.
Visit this group at http://groups.google.com/group/browserscope.
For more options, visit https://groups.google.com/d/optout.

Steve Souders

unread,
Jul 15, 2014, 5:19:04 PM7/15/14
to browse...@googlegroups.com
I guess it's debatable whether this is a bug. 

<a href="http://msdn.microsoft.com/en-us/library/ie/dn265039(v=vs.85).aspx">IE Dev Center</a> says prefetch is only supported on "a top level document; that is, it can't be an iframe document". But <a href="http://www.w3.org/TR/html5/links.html#link-type-prefetch">the spec</a> doesn't dictate this limitation.

There are examples in the real world where people are using link rel prefetch as a response header in resources *other than* the top level document, such as sapato.ru and lensaindonesia.com. These sites are #14K and #68K worldwide, so they're not huge but not minor either. I think there's a huge benefit in being able to specify resources to prefetch in responses other than the top level document.

Perhaps it's good that the current test uses iframes as that's closer to the spec.

-Steve


On Tuesday, July 15, 2014 1:52:33 PM UTC-7, Lindsey Simon wrote:
On Mon, Jul 14, 2014 at 8:06 AM, Jonathan Klein <jonathan...@gmail.com> wrote:
Hello,

Browserscope reports that IE11 doesn't support link rel="prefetch", but it does, just not in subframes.  See the discussion here: https://twitter.com/ericlaw/status/488700258373758976.  Is it possible to modify the test to expose this distinction?  

Actually, I don't believe so since the test harness is done with frames. =(
 

Cheers,

Jonathan

--
You received this message because you are subscribed to the Google Groups "Browserscope" group.
To unsubscribe from this group and stop receiving emails from it, send an email to browserscope+unsubscribe@googlegroups.com.

Jonathan Klein

unread,
Jul 16, 2014, 9:41:25 AM7/16/14
to browse...@googlegroups.com
I agree that there is value in showing whether the browser supports prefetching from iframes, but in my mind the ideal would be to have both test cases, or at least have some way to indicate that IE11 supports them at the top level document (this could even be a hardcoded text note about it).  I consider browserscope to be the truth for browser support, and as a result I was under the impression that IE11 didn't have any support for link rel="prefetch" for quite a while.  It would be great if we could find a way to stop other people from making this same mistake.  


To unsubscribe from this group and stop receiving emails from it, send an email to browserscope...@googlegroups.com.

Steve Souders

unread,
Jul 16, 2014, 2:55:56 PM7/16/14
to browse...@googlegroups.com
Totally agree - as I said it'd be great to test BOTH cases. Alternatively we could add a hardcoded footnote. But I don't a way to do either of those that isn't a lot of work. Lindsey - can you think of a hack to add a footnote? The hard part of that is there are so many sliced views of the data. 

I don't think it's worth doing a lot of work to clarify a browser's partial support of a feature that violates the spec. Eric Lawrence filed a bug with IE so. If there's an easy fix I'd support that. Ow, I think it's better to urge IE to support the spec.

-Steve


On Wednesday, July 16, 2014 6:41:25 AM UTC-7, Jonathan Klein wrote:
I agree that there is value in showing whether the browser supports prefetching from iframes, but in my mind the ideal would be to have both test cases, or at least have some way to indicate that IE11 supports them at the top level document (this could even be a hardcoded text note about it).  I consider browserscope to be the truth for browser support, and as a result I was under the impression that IE11 didn't have any support for link rel="prefetch" for quite a while.  It would be great if we could find a way to stop other people from making this same mistake.  

Lindsey Simon

unread,
Jul 17, 2014, 1:39:31 AM7/17/14
to browse...@googlegroups.com
On Wed, Jul 16, 2014 at 11:55 AM, Steve Souders <steveso...@gmail.com> wrote:
Totally agree - as I said it'd be great to test BOTH cases. Alternatively we could add a hardcoded footnote. But I don't a way to do either of those that isn't a lot of work. Lindsey - can you think of a hack to add a footnote?

Yeah, I mean we could statically add a note to the test page, but it would be less easy to do it where it matters (the result table).
 
The hard part of that is there are so many sliced views of the data. 


Yeah, this ;)

 
I don't think it's worth doing a lot of work to clarify a browser's partial support of a feature that violates the spec.

Handy, because I don't think it's a simple amount of work anyway to try to detect differences with and without frames or annotate particular tests where it matters
-l
 
Eric Lawrence filed a bug with IE so. If there's an easy fix I'd support that. Ow, I think it's better to urge IE to support the spec.

-Steve


On Wednesday, July 16, 2014 6:41:25 AM UTC-7, Jonathan Klein wrote:
I agree that there is value in showing whether the browser supports prefetching from iframes, but in my mind the ideal would be to have both test cases, or at least have some way to indicate that IE11 supports them at the top level document (this could even be a hardcoded text note about it).  I consider browserscope to be the truth for browser support, and as a result I was under the impression that IE11 didn't have any support for link rel="prefetch" for quite a while.  It would be great if we could find a way to stop other people from making this same mistake.  


On Tue, Jul 15, 2014 at 5:19 PM, Steve Souders wrote:

I guess it's debatable whether this is a bug. 

<a href="http://msdn.microsoft.com/en-us/library/ie/dn265039(v=vs.85).aspx">IE Dev Center</a> says prefetch is only supported on "a top level document; that is, it can't be an iframe document". But <a href="http://www.w3.org/TR/html5/links.html#link-type-prefetch">the spec</a> doesn't dictate this limitation.

There are examples in the real world where people are using link rel prefetch as a response header in resources *other than* the top level document, such as sapato.ru and lensaindonesia.com. These sites are #14K and #68K worldwide, so they're not huge but not minor either. I think there's a huge benefit in being able to specify resources to prefetch in responses other than the top level document.

Perhaps it's good that the current test uses iframes as that's closer to the spec.

-Steve


On Tuesday, July 15, 2014 1:52:33 PM UTC-7, Lindsey Simon wrote:



On Mon, Jul 14, 2014 at 8:06 AM, Jonathan Klein <jonathan...@gmail.com> wrote:
Hello,

Browserscope reports that IE11 doesn't support link rel="prefetch", but it does, just not in subframes.  See the discussion here: https://twitter.com/ericlaw/status/488700258373758976.  Is it possible to modify the test to expose this distinction?  

Actually, I don't believe so since the test harness is done with frames. =(
 

Cheers,

Jonathan



Jonathan Klein

unread,
Jul 22, 2014, 9:58:18 AM7/22/14
to browse...@googlegroups.com
Thanks for the follow-up guys, it sounds like waiting for the IE team to fix this is the best approach, and in the meantime we can communicate this difference in support via our respective channels.  

Cheers,

Jonathan

davi...@outlook.com

unread,
Nov 25, 2014, 1:20:47 PM11/25/14
to browse...@googlegroups.com
Apologies for being late to the discussion.  IE is working as designed.  We had a hard limit of 10 prefetches and we decided it's better to give all 10 of the prefetches to the root page and not have arbitrary iframes (e.g. ads) eat into the prefetch quota.

cheers,
_dave_
Member of the IE team.
Reply all
Reply to author
Forward
0 new messages