Posted by Marcus Bointon on 05/11/05 01:48

On 10 May 2005, at 20:19, Vicente Werner wrote:

> until they put
> a really stable javascript implementation, javascript is nothing but
> trouble:

I did say that I'd be happy with IE6 only, which has remained
unchanged for nearly 4 years. How much more stable do you want it?

> Take for example qforms, probably the most robust dhtml
> validation system I know of (better than formcat, formsess, etc..),
> still has a significant amount of browsers where it dosn't work as it
> should, and even on those that looks like works flawlessly there're
> some situations that ruin it.

With that defeatist attitude you can never use JS for anything. IE6
doesn't conform to some W3C specs - looks like we'll have to scrap
HTML and CSS too. Firefox fails Acid2? Better dump that as well.
You're using gmail, which is about as complex as JS gets, yet it
works quite nicely in a large proportion of browsers. So google
should can it because it doesn't all work in lynx? JS for validation
is a walk in the park by comparison. If I can do client-side
validation reliably in IE6, which is used by 90% of my visitors,
there is no reason to deny it to them because some small proportion
of the remaining 10% might have trouble, especially as it's trivial
to disable it altogether for them (Except for the compulsory idiot
running Opera3 with an IE6 agent string).

> Call me lazy, but certainly I don't see the benefits outweighting the
> effort just to make it run, you still have to duplicate work and it's
> not worth it.

I think you're just not getting what I'm on about. What I'm asking
for is to write ONE piece of code that's probably quite similar to
what you'd already do with SmartyValidate, and have it render output
for both client and server. It's doing double the work for the same
amount of coding effort. Sure, it needs a well-tested JS library to
build on (qforms would do nicely), but once bundled into a nice
Smarty extension, the API should be simple. I agree that coding it
all manually twice is a silly idea - that's why I started this thread.

> Qforms [...] is mostly a
> GOOD validation tool.

But from what you've been saying, it's no use at all because
somewhere there may be one browser that it doesn't work in, even if
you only explicitly enable it in those that do.

PEAR's HTML_QuickForm does look pretty good, and does do what I'm
after, though it's quite heavy going.

Marcus
--
Marcus Bointon
Synchromedia Limited: Putting you in the picture
marcus@synchromedia.co.uk | http://www.synchromedia.co.uk

[Back to original message]