Re: [PEPr] Proposal for Images::Image_CAPTCHA

Christian Wenz wrote:

Philippe Jausions wrote:

This should be an interesting and useful package. Please have a look at Image_Imagick (PEPr) and Image_Transform packages for possible compatibility/interoperability and/or dependency.

I will most probably check this over the weekend, especially Image_Transform!

What's missing for now in Image_Transform is a filter-like capability, which would be very usefull in your case... Image_Imagick is on the other side more capable of filter-based processing, but only with the imagick PECL extension... but hopefully everything could be tie together to be even more powerful. My comment was more to keep an eye on all these packages to possibly make them work together in the not too distant future.

It would also be nice to not solely rely on Text_Password for text generation. One may want to provide the text to be added.

That's a good idea, I'll create a way to provide a custom phrase in the class' constructor.

Would it be safe to assume that you can leave it to the application itself to verify the answer to the "CAPTCHA image" instead of relying only on a built-in method?... I haven't looked in the code yet (:-p), but do you allow concurrent "pictures" within one session (users may have several web browser windows opened at the same time...)

Also, provide an interface flexible enough to also be able to use pictures instead of text as patterns to recognize.

so you mean there is an apple and an orange in the image and the user has to type "apple orange"? Or didn't I get that right?

Mutliple items may be ambigous, but yes this is the idea... For instance, the question could be "What is this animal?" and then the picture could be of a "dog", "cat", "cow"... whatever... I guess you get the idea. The one using this method should of course pick non-ambigous pictures but that's not our problem... -Philippe