At 09:52 AM 6/26/2003 +0200, you wrote:
On Èt, 2003-06-26 at 09:36, Jacob
Kjome wrote:
> XMLC-2.2 is close to ready. I wanted to check on one thing,
though. In
> OutputOptions, the javadoc for getUseAposEntity() said that the
default was
> false. However, the field actually defaulted to true. I
changed it to
> false, but I wonder if the default shouldn't actually be true since
it is
> actually a quirk that some browsers might not be able to deal with
'
> and XML requires the escaping of single quotes since they can be
used to
> surround attributes.
Well, there was the problem with _javascript_ in XHTML. XMLC converted
the
quotes of _javascript_ function parameters to ' and browsers
didn't
understand that. The solution was to replace ' with some hex
code,
wasn't it? I thought you had done that already.
What I had done, actually, is leave the existing work around in place,
which is get/setUseAposEntity(). The work around using the numeric
character escape sequence which I mentioned on the list wasn't well liked
by some of the XMLC committers. Said methods work for the
situation. I just figured they were unnecessary if we could use the
numeric escape, but the methods will work fine as well.
The question is if the
hex code in the _javascript_ parameters is
understood by browsers correctly or not.
Actually, the question is whether we should use the quirk of not escaping
the single quotes as a default when that violates the XML spec (for the
case where you are outputting XML or XHTML). What I am talking
about is making the default for escaping single characters with
' be "true" instead of "false" which I
switched it to, but am now having second thoughts. One can always
turn this off by using OutputOptions#setUseAposEntity(false) and you will
be golden. This would make XMLC consistent in how it uses
quirks. No Qurik would be used by default. Only via a custom
OutputOptions would they be used. The defaults would be entirely
non-quirks oriented.
Does that make more sense?
Jake
| 