[ZPT] ZPT and TAL 1.0 Final Release

[Guido van Rossum]

> From: "Guido van Rossum" <guido@digicool.com>
> > It's easy enough to skip the parsing when there's no attrribute
> > replacement.  But if we don't parse, we can't be sure that the
> > inserted text doesn't add unbalanced elements, and this means that a
> > mistake in the inserted text could screw up the page globally (rather
> > than just locally).
> 
> I'd worry about this in XML, but not in HTML.  If you generate HTML with
> DTML, you take your chances, and if you want structural integrity you use
> ZPT to generate it instead.

I'm not sure I like this, but it would certainly make things a lot
easier.  Then "structure" just means "don't quote < > or &".

> > When there *is* attribute replacement, I could implement the attribute
> > replacement without using a full-blown parser, but it would be
> > hackish.
> 
> I have yet to use tal:replace and tal:attributes on the same tag, so this
> may be a non-issue.

I've wanted to drop this bit of the spec all along, knowing that it
would be a pain to implement, but other forces (you? Jim?) requested
it.  I'd happily drop it.

> > Then we should define what makes something a TAL program.  A list
> > whose first item is the tuple ("version", "1.0")???  I think we should
> > do something better.  I guess I can design a better starting item for
> > ZPT and TAL 1.1.
> 
> Define a TAL program class, and test isinstance?

Maybe, although that's not how it's done now.  Maybe it should be.

--Guido van Rossum (home page: http://www.python.org/~guido/)