IE 8 loses last tag(s) and displays </ at bottom of page

Hi everyone, I have a java web app that generates HTML pages and sends them
to IE 8 clients.

The HTML appears to be well formed and includes all required ending tags and
I have verified that the entire HTML stream is flushed to the client and that
the proper content length is set in the response headers.

On some pages, the client page seems to cut off somewhere near the end of
the HTML (sometimes within a hidden input definition or usually within the
trailing </html> tag.

The view source shows me that the end of the html is missing which explains
why the </ appears at the end of the page.

It does not happen on every page and I have been unable to determine the
pattern that makes it happen.

Any ideas on what may be causing this or where to look will be greatly
appreciated.

Cheers

Martin Kultermann
eStar Communications

"Martin Kultermann" <MartinKultermann@discussions.microsoft.com> wrote in
message news:1426A9E7-F0AB-4450-A97A-4FC7433B1DA8@microsoft.com...
> Hi everyone, I have a java web app that generates HTML pages and sends
> them
> to IE 8 clients.
>
> The HTML appears to be well formed and includes all required ending tags
> and
> I have verified that the entire HTML stream is flushed to the client and
> that
> the proper content length is set in the response headers.
>
> On some pages, the client page seems to cut off somewhere near the end of
> the HTML (sometimes within a hidden input definition or usually within the
> trailing </html> tag.
>
> The view source shows me that the end of the html is missing which
> explains
> why the </ appears at the end of the page.
>
> It does not happen on every page and I have been unable to determine the
> pattern that makes it happen.
>
> Any ideas on what may be causing this or where to look will be greatly
> appreciated.
>
> Cheers
>
> Martin Kultermann
> eStar Communications

You're going to have to look at tracing the underlying data packets and see
if there's a problem there, IE will just render what is has been sent. I'm
guessing that the Content-Length will be off slightly - you may think it's
correct, but if you're sending any unicode characters and not taking into
account double bytes, for instance, you could be a few bytes off on your
length calculation.

Try something like Fiddler2 http://www.fiddler2.com/fiddler2/ and look at
the response data headers and content for these requests.

--
Dan

Developer-specific resources include:

MSDN IE Development Forum (post such questions here instead)
http://social.msdn.microsoft.com/For...opment/threads

Tip: When posting in Developer forums, always include a link to your web
site or test pages in your first post.

IE Developer Center
http://msdn.microsoft.com/en-us/ie/default.aspx

Learn IE8
http://msdn.microsoft.com/en-us/ie/aa740473.aspx

HTML and DHTML Overviews and Tutorials
http://msdn.microsoft.com/en-us/library/ms537623.aspx and

Cascading Style Sheets (CSS)
http://msdn2.microsoft.com/en-us/ie/aa740476.aspx

Expression Web SuperPreview for Internet Explorer (free, stand-alone visual
debugging tool for IE6, IE7, and IE8)
http://www.microsoft.com/downloads/d...b-dccff3fae677

Expression Web SuperPreview Release Notes
http://www.microsoft.com/expression/...easeNotes.aspx

Validators:
http://validator.w3.org/
http://jigsaw.w3.org/css-validator/


Martin Kultermann wrote:
> Hi everyone, I have a java web app that generates HTML pages and sends
> them
> to IE 8 clients.
>
> The HTML appears to be well formed and includes all required ending tags
> and
> I have verified that the entire HTML stream is flushed to the client and
> that the proper content length is set in the response headers.
>
> On some pages, the client page seems to cut off somewhere near the end of
> the HTML (sometimes within a hidden input definition or usually within the
> trailing </html> tag.
>
> The view source shows me that the end of the html is missing which
> explains
> why the </ appears at the end of the page.
>
> It does not happen on every page and I have been unable to determine the
> pattern that makes it happen.
>
> Any ideas on what may be causing this or where to look will be greatly
> appreciated.
>
> Cheers
>
> Martin Kultermann
> eStar Communications

Thanks a lot Dan, I think this is solved.

I was writing the client response stream to a file as well to help with the
diagnosis.
The length of that file reported by Windows, the Content Length that I
placed into the response header and the body size reported by Fiddler are all
exactly the same.

Then I tried removing the Content Length header to see if that makes a
difference and it did, it now seems to work.

I was under the impression that performance would suffer if you did not
specify a content length, because the client would not know when the stream
is done or might wait for more. Do you know if that's true?

"Dan" wrote:

>
> "Martin Kultermann" <MartinKultermann@discussions.microsoft.com> wrote in
> message news:1426A9E7-F0AB-4450-A97A-4FC7433B1DA8@microsoft.com...
> > Hi everyone, I have a java web app that generates HTML pages and sends
> > them
> > to IE 8 clients.
> >
> > The HTML appears to be well formed and includes all required ending tags
> > and
> > I have verified that the entire HTML stream is flushed to the client and
> > that
> > the proper content length is set in the response headers.
> >
> > On some pages, the client page seems to cut off somewhere near the end of
> > the HTML (sometimes within a hidden input definition or usually within the
> > trailing </html> tag.
> >
> > The view source shows me that the end of the html is missing which
> > explains
> > why the </ appears at the end of the page.
> >
> > It does not happen on every page and I have been unable to determine the
> > pattern that makes it happen.
> >
> > Any ideas on what may be causing this or where to look will be greatly
> > appreciated.
> >
> > Cheers
> >
> > Martin Kultermann
> > eStar Communications
>
> You're going to have to look at tracing the underlying data packets and see
> if there's a problem there, IE will just render what is has been sent. I'm
> guessing that the Content-Length will be off slightly - you may think it's
> correct, but if you're sending any unicode characters and not taking into
> account double bytes, for instance, you could be a few bytes off on your
> length calculation.
>
> Try something like Fiddler2 http://www.fiddler2.com/fiddler2/ and look at
> the response data headers and content for these requests.
>
> --
> Dan
>
> .
>

It depends on the web server, and if it's a true stream or just buffered
content. If it's buffered, most web servers will add the Content-Length
header themselves. If there's no header, the browser will just accept the
data until the connection is closed, so this shouldn't cause you any
problems unless your application never closes the TCP socket after sending
the data.

Dan


"Martin Kultermann" <MartinKultermann@discussions.microsoft.com> wrote in
message news:676D9AF0-3A74-4882-9C72-A4B32A124E3D@microsoft.com...
> Thanks a lot Dan, I think this is solved.
>
> I was writing the client response stream to a file as well to help with
> the
> diagnosis.
> The length of that file reported by Windows, the Content Length that I
> placed into the response header and the body size reported by Fiddler are
> all
> exactly the same.
>
> Then I tried removing the Content Length header to see if that makes a
> difference and it did, it now seems to work.
>
> I was under the impression that performance would suffer if you did not
> specify a content length, because the client would not know when the
> stream
> is done or might wait for more. Do you know if that's true?
>
> "Dan" wrote:
>
>>
>> "Martin Kultermann" <MartinKultermann@discussions.microsoft.com> wrote in
>> message news:1426A9E7-F0AB-4450-A97A-4FC7433B1DA8@microsoft.com...
>> > Hi everyone, I have a java web app that generates HTML pages and sends
>> > them
>> > to IE 8 clients.
>> >
>> > The HTML appears to be well formed and includes all required ending
>> > tags
>> > and
>> > I have verified that the entire HTML stream is flushed to the client
>> > and
>> > that
>> > the proper content length is set in the response headers.
>> >
>> > On some pages, the client page seems to cut off somewhere near the end
>> > of
>> > the HTML (sometimes within a hidden input definition or usually within
>> > the
>> > trailing </html> tag.
>> >
>> > The view source shows me that the end of the html is missing which
>> > explains
>> > why the </ appears at the end of the page.
>> >
>> > It does not happen on every page and I have been unable to determine
>> > the
>> > pattern that makes it happen.
>> >
>> > Any ideas on what may be causing this or where to look will be greatly
>> > appreciated.
>> >
>> > Cheers
>> >
>> > Martin Kultermann
>> > eStar Communications
>>
>> You're going to have to look at tracing the underlying data packets and
>> see
>> if there's a problem there, IE will just render what is has been sent.
>> I'm
>> guessing that the Content-Length will be off slightly - you may think
>> it's
>> correct, but if you're sending any unicode characters and not taking into
>> account double bytes, for instance, you could be a few bytes off on your
>> length calculation.
>>
>> Try something like Fiddler2 http://www.fiddler2.com/fiddler2/ and look
>> at
>> the response data headers and content for these requests.
>>
>> --
>> Dan
>>
>> .
>>