Re: [cgiapp] RFC: CGI::Application::Plugin::CAPTCHA

["Jason A. Crome" <suppressed>]

On Aug 25, 2005, at 12:38 PM, Michael Peters wrote:

> I'm not saying evey runmode should call create_captcha(), I'm saying
> create_captcha() should be a run mode in and of itself.

Thanks! :)  That was painfully obvious when you explained it to me.   
I can't believe that I missed that before ;)

> > So, essentially we change the type of header produced by C::A, then
> > return the image data from the runmode?  Am I following correctly?
>
> Correct. The header should include the cookie information and the
> content type (like 'image/gif', etc).
>
> > If
> > so, what of cgiapp_postrun?  For those of us that sometimes alter
> > headers there, this would be an issue to be contended with.
>
> That's a good question. It's possible that a postrun could alter  
> headers
> incorrectly or mangle the image. There's a couple of options that come
> to mind quickly. How good they are is definitely a wildcard :)
>
> + Check your get_current_rm() in postrun and don't do anything if it's
> 'create_captcha'. This could be a headache if you do postrun  
> processing
> in your base class (but not too bad). The biggest issue would be other
> plugins trying to do something there.

I think I like this the best of the three.  I don't think that it's  
any more strange than my multiple inheritance requirement for  
CAP::HTDot.  I would hope that, in most cases, it would be a small  
tradeoff to get the CAPTCHA functionality.

> + Have a separate class that just does the CAPTCH generation. This  
> would
> of course need it's one instance script (if you use them). It would be
> separate from the code using the CAPTCHA, but it shouldn't matter. But
> it does seem sorta hackish

Hackish, but it works ;-)

> + Maybe C::A needs a way for a plugin to not only register a callback,
> but also unregister a hook/callback. So the create_captcha() run mode
> would just make sure there are no postruns running. This could cause
> problems though if you have postruns that need to run that wouldn't
> affect the headers or the image.

Interesting, but I see a lot of potential breakage as a result of that.

Thanks for the ideas, explanation, and help :)
Jason


---------------------------------------------------------------------
Web Archive:  http://www.mail-archive.com/suppressed/
             http://marc.theaimsgroup.com/?l=cgiapp&r=1&w=2
To unsubscribe, e-mail: suppressed
For additional commands, e-mail: suppressed