On Thu, 5 Jul 2007, Tobias Schlitt wrote:

> On 07/05/2007 10:27 AM Derick Rethans wrote:
> > On Wed, 4 Jul 2007, Alexandru Stanoi wrote:
> 
> >> Flexibility is the key here, so I propose this:
> >>
> >> validateEmailAddress - just regexp
> >> verifyDomain - just checkdnsrr() or exec( 'nslookup' ) on Windows
> >> convertToPunyCode - converts to Punycode
> >> convertToUnicode - viceversa
> >>
> >> and it will be up to the developer to use these functions as he sees fit.
> 
> > Yes, this idea sounds fine, but we need more descriptive method names. I 
> > would also make the rx checks as option to validateEmailAddress 
> > actually, as it's the same function and that shouldn't be done by two 
> > methods. I think that we should also see if we can do more than just 
> > "checkdnsrr()" (you mentioned some more advanced algorithm).
> 
> I think the sense was, that options to a function are not good?

It's not bad just because it's an option...

> Anyway, I dislike it here especially, since we force users to user our 
> regex, if they just want to perform our mx checks. It should be 2 
> seperate methods.

That makes sense though.

> > The last two functions "convertToPunyCode" and "convertToUnicode" do 
> > actually not really belong in the Mail component at all, as they are 
> > generic methods. I think they actually belong more in Url. I would 
> > therefore suggest to create two methods on ezcUrl (and the algorithms it 
> > self in ezcUrlTools as static methods) that can convert a Url toASCII()
> > (with the punycode algorithms) and toUnicode() with the reverse 
> > algorithm. (I am naming them toASCII() and toUnicode() here because 
> > that's what wikipedia suggests). 
> 
> That would mean a dependency Mail -> Url?

No, why? AFAIK we basically agreed that we don't do any of those 
conversions ourselves, but that's up to the application.

Derick
-- 
Components mailing list
Components@lists.ez.no
http://lists.ez.no/mailman/listinfo/components

Reply via email to