[10317] in www-talk@info.cern.ch
Re: Simple(?) comments question...
daemon@ATHENA.MIT.EDU (S.N.Brodie@ecs.soton.ac.uk)
Thu Sep 19 07:11:11 1996
From: S.N.Brodie@ecs.soton.ac.uk
To: galactus@htmlhelp.com (Arnoud "Galactus" Engelfriet)
Date: Thu, 19 Sep 1996 12:06:10 +0100 (BST)
Cc: www-talk@w3.org
In-Reply-To: <SVEQy4uYOtCS089yn@htmlhelp.com> from "Arnoud "Galactus" Engelfriet" at Sep 18, 96 08:54:10 pm
Resent-From: www-talk@w3.org
Arnoud "Galactus" Engelfriet wrote:
>
> In article <828.9609180920@strachey.ecs.soton.ac.uk>,
> S.N.Brodie@ecs.soton.ac.uk wrote:
> > My impression is that this is not a correctly terminated comment, since
> > it does not fit the strict definition given in RFC1866. However,
> > that's irrelevant, as all 3 browsers I've tried it on accept it.
>
> I suppose these browsers simply consider "<!--" the comment starting
> tag and "-->" the corresponding closing tag. My favourite way to
> demonstrate that is the following *legal* comment:
>
> <!-- -- --> -->
>
> That's two comments, one of which only contains " " and one contains
> "> ".
Agreed.
> Anyway, as far as I can see RFC 1866 does not discuss the "-" character
> as last character before "--" explicitly. It only states (section 3.2.5)
>
> Each comment starts with `--' and includes
> all text up to and including the next occurrence of `--'.
>
> I'm just confused if the sequence "---" _should_ be seen as "-"
> followed by "--" or as "--" followed by "-".
That is the problem. My (very recently modifed :-) parser accepts the
following and displays "Body text." as it should:
Body <!-- comment -- -- > shouldn't see this! --> text.
Netscape gets itself into all kinds of a mess with this. It seems to be
applying the "-->" terminates a comment and if we don't find one, go back
to the first occurrence of a > For example:
One <!-- hi -- -- > lo --> Two
is displayed as "One Two". However:
One <!-- hi -- -- > lo -- > Two
is displayed as "One lo -- > Two"
Anybody got any suggestions how --- should be parsed whilst parsing a
comment structure? My inclination is to treat it differently depending
on whether you are inside a comment or not, but it becomes a special
case that way, since behaviour will have to change depending on the
number of consecutive - characters. Obviously you have to keep track
of whether you are "inside" a -- or not. Having seen <!-- parser goes
into comment parsing mode, and sets a flag "in_double_dash" to true.
Then it continues until it sees 2 or more consecutive - characters. It
discards all of these characters and sets "in_double_dash" to false. A
'>' is only accepted as terminator if in_double_dash is false. Upon
seeing a -- whilst in_double_dash is zero, set in_double_dash to one.
Whilst this has a set behaviour for 2+ consecutive - symbols, is it the
desired behaviour?
One of the places this is most likely to crop up (IMHO) is when reading
inlined scripts. I have specifically added recognition for <script>
and </script> to the parser so it can discard that stuff automatically
without relying on the script being commented out. I discard it as I
have no intention of porting Visual Basic or writing a Javascript
interpreter. I just don't have the time.
--
Stewart Brodie, Electronics & Computer Science, Southampton University.
http://www.ecs.soton.ac.uk/~snb94r/ http://delenn.ecs.soton.ac.uk/
