Comments on: Writing a face recognition library interface is hard. Really hard. http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/ { on programming and the internets, every monday } Mon, 23 Aug 2010 09:39:45 +0000 http://wordpress.org/?v=2.9.1 hourly 1 By: tangential http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10412 tangential Wed, 10 Jun 2009 03:33:00 +0000 http://lbrandy.com/blog/?p=817#comment-10412 As a Qt developer, thankyou ;) It feels good to be appreciated, and we're also all pretty passionate here about writing good libraries. May I reccommend some mandatory reading for anyone working here at Qt ;) http://chaos.troll.no/~shausman/api-design/api-design.pdf (If that link fails, google "The Little Manual of API Design" by Jasmine Blanchette). As a Qt developer, thankyou ;) It feels good to be appreciated, and we’re also all pretty passionate here about writing good libraries.

May I reccommend some mandatory reading for anyone working here at Qt ;) http://chaos.troll.no/~shausman/api-design/api-design.pdf (If that link fails, google “The Little Manual of API Design” by Jasmine Blanchette).

]]> By: Anonymous http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10388 Anonymous Tue, 09 Jun 2009 13:59:55 +0000 http://lbrandy.com/blog/?p=817#comment-10388 How fun that while I was reading the summary my thoughts immediately went to Qt before you mentioned it :) Kudos to all Qt fans. How fun that while I was reading the summary my thoughts immediately went to Qt before you mentioned it :) Kudos to all Qt fans.

]]> By: Monkeyget http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10385 Monkeyget Tue, 09 Jun 2009 11:03:40 +0000 http://lbrandy.com/blog/?p=817#comment-10385 There is a nice presentation which put the emphasis on the various levels of integration of an API. Designing and Evaluating Reusable Components : http://mollyrocket.com/9438 There is a nice presentation which put the emphasis on the various levels of integration of an API. Designing and Evaluating Reusable Components : http://mollyrocket.com/9438

]]> By: Ollie Saunders http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10374 Ollie Saunders Tue, 09 Jun 2009 01:36:44 +0000 http://lbrandy.com/blog/?p=817#comment-10374 Continuously and consistently apply your solution back to the problem domain. In short, test your API early and often in the most diverse and real-world ways you can. Continuously and consistently apply your solution back to the problem domain. In short, test your API early and often in the most diverse and real-world ways you can.

]]> By: Michael http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10368 Michael Mon, 08 Jun 2009 22:24:33 +0000 http://lbrandy.com/blog/?p=817#comment-10368 Also there's a one hour Google Tech Talk on YouTube titled "How To Design A Good API and Why it Matters": http://www.youtube.com/watch?v=aAb7hSCtvGw Also there’s a one hour Google Tech Talk on YouTube titled “How To Design A Good API and Why it Matters”: http://www.youtube.com/watch?v=aAb7hSCtvGw

]]> By: Michael http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10367 Michael Mon, 08 Jun 2009 22:16:45 +0000 http://lbrandy.com/blog/?p=817#comment-10367 I really liked this post. I was a little surprised when you said that the face-face scoring was low-level. To me that seems pretty high level! A key aspect of programming is breaking down problems into manageable chunks. The Unix philosophy was to combine a lot of little programs that perform their little tasks really well, rather than writing a few programs that did everything which would end up being mediocre. Your API could be implemented in several layers/components, each with a specific purpose, and the user can choose what level of abstraction he wants to work at. It seems to me that use cases are very important for this--they lead to real-world test applications, as Tony mentioned. If you were to expose only the minimal functionality to achieve your use cases, I wonder where that would take you... I really liked this post. I was a little surprised when you said that the face-face scoring was low-level. To me that seems pretty high level!

A key aspect of programming is breaking down problems into manageable chunks. The Unix philosophy was to combine a lot of little programs that perform their little tasks really well, rather than writing a few programs that did everything which would end up being mediocre. Your API could be implemented in several layers/components, each with a specific purpose, and the user can choose what level of abstraction he wants to work at.

It seems to me that use cases are very important for this–they lead to real-world test applications, as Tony mentioned. If you were to expose only the minimal functionality to achieve your use cases, I wonder where that would take you…

]]> By: Tony Arkles http://lbrandy.com/blog/2009/06/writing-a-face-recognition-library-interface-is-hard/comment-page-1/#comment-10362 Tony Arkles Mon, 08 Jun 2009 18:43:14 +0000 http://lbrandy.com/blog/?p=817#comment-10362 Another good way to hammer out a good API is to make a first cut at it, and then spend a few days writing sample applications (beyond "hello world") with the library. Get them to report back the parts that were awkward to work with. Iterate, fix the biggest few complaints, and do it again. You should hit a pretty decent API within a few goes at it. Another good way to hammer out a good API is to make a first cut at it, and then spend a few days writing sample applications (beyond “hello world”) with the library. Get them to report back the parts that were awkward to work with. Iterate, fix the biggest few complaints, and do it again. You should hit a pretty decent API within a few goes at it.

]]>