> -----Original Message-----
> From: Clay Leeds [mailto:[EMAIL PROTECTED]
> On Dec 24, 2003, at 8:55 AM, Bernd Brandstetter wrote:
> >
> > wouldn't it be reasonable to also accept the obsolete 'master-name'
> > and to
> > just print out a corresponding warning message instead of throwing an
> > error? AFAIK, the attribute has only been renamed to
> > 'master-reference',
> > but its meaning and usage remained identically the same.
> >
>
> +1 (if non-votes count! ;-) )
>
> In fact, I would love to see this considered a "bugfix" for the 0.20.5
> maintenance release, as I believe it would help people upgrade from
> 0.20.4 (or lower) to 0.20.5 and higher. Backward-compatibility is
> always a nice thing--especially if it's as "easy" as this appears to
> be.
>

Yes, perhaps some solution comparable to deprecation in Java. As ( and if )
the spec evolves further, this would become a necessity anyway. ( Problem
would appear if, in the future 'master-name' would be defined in a whole
different way... XSL-FO doesn't provide a 'version' attribute yet, like XSLT
does )

Then again, it seems hard to believe this topic has not presented itself
before, and maybe it wasn't so easy after all.

Since one of the primary goals of FOP is compliance with the spec, and if
the spec no longer considers this property name to be valid --what's more
important? Backward compatibility or compliance with (the most current
version of) the spec?

The latter seems preferrable as it discourages the use of properties that
are undefined by the spec, while the first allows people to ignore it. If at
some time they decide (for some mysterious reason) to use XEP instead of
FOP, they'll receive an error anyway.

Just a thought.


Cheers,

Andreas

Reply via email to