[l2h] Re: Weirdness for MSIE5 under NT4 SP4

Fred L. Drake, Jr. Fred L. Drake, Jr." <fdrake@acm.org
Sat, 13 Nov 1999 17:10:41 -0500 (EST)

Ross Moore writes:
 > Fred, can you please give URLs to pages that you know to be affected?
 > The comment is the one that starts:


was the example mentioned by the person reporting the problem (I don't 
have the right service pack installed on my NT box).  Note that the
problem *only* has been confirmed for IE5 + NT4 + SP4; service packs 3 
& 5 don't exhibit this problem; no report regarding the new SP6.  Its
also only on the reload of the page, not the initial load.

 > Yes, the comment is still generated in that position.
 > though now it is preceeded by a blank line.

  I'll ask the person reporting the error to test this.

 > I believe it to be perfectly valid according to SGML rules,
 > as well as HTML --- don't know about XML.

  I think it's legal in all these cases.  The problem appears highly
specific to the environment mentioned; other products which use the
same HTML parsing DLL don't seem to exhibit the problem at all.

 > Moving the comment does not solve the problem for the tens of thousands
 > of files already created with LaTeX2HTML, whereas a bug-fix from Microsoft
 > would be a complete solution; so I'm inclined to just ignore this one.
 > Besides, moving that information comment to after <HTML>
 > can mean putting it a long way down the file, when
 > there are lots of <META> tags in the header portion.
 > It just doesn't seem appropriate to do this, sorry.
 > On the other hand, if there is an error in the syntax of the comment
 > itself, then that should indeed be fixed. But then why is it just being
 > discovered now? 

  The comment syntax is correct.  I think the software combination is
the issue, and since it only seems to occur on a user requested
reload, it may be very difficult to detect the first time.

 > But wait-on.
 > I've seen similar behaviour before, which I rationalised as follows.
 > THe 'Refresh' button could be reloading the last file down-loaded.
 > This is often *not* the HTML page that the browser is displaying,
 > but the CSS stylesheet used in determining how to display the information
 > content of the HTML file.
 > It used to be a real problem with Netscape, when an HTML file asked
 > for a CSS stylesheet, but that file could not be found.
 > Netscape threw up an error and would not display the HTML document.
 > Trying to back-out using the `Back` button would not work,
 > since this would get the HTML file, which would ask for the same
 > missing CSS style-sheet.
 > Maybe there is a similar problem with MS's Refresh that is asking
 > to display the CSS as if it were HTML. This would probably result
 > in a blank page. Why the position of the comment makes any difference;
 > well that's anyone's guess.

  I'll follow up on this with the reporter of the problem, to see if
this is a possibility here.
  Thanks for the response!


Fred L. Drake, Jr.	     <fdrake@acm.org>
Corporation for National Research Initiatives