Page 1 of 1

Any New Insights

Posted: Mon Aug 29, 2011 7:24 am
by adam33147
Wondering if anyone has any new insights they came upon. Anything that is cool to you. Would love to hear about them. Or even if you dont have any new insights, how about your domains of interest? Your technical musings. What would you like to be able to do? What are you working on? What are you curious about implementing.

For example. I would love to be able to implement https. But Im not sure how to get into "hacker mode" on this one yet.

A cool idea I have is to implemnt a program "Autoclient" that can automate datamining. It can be a webclient wrapped in a function that can make specified requests, but for example could hold a session with a server. This could be wrapped in a function that can dig the data from the servers response, so calling this from say mapcar could give you the mined data according to a list of references.

Whatever inspires, please feel free to post here.

Re: Any New Insights

Posted: Mon Aug 29, 2011 11:00 am
by Duke
adam33147 wrote:Whatever inspires, please feel free to post here.
I'm currently procrastinating on implementing a lot of stuff that already exists in SDL. I've already done lines, circles, and polygons, then I moved on to reinventing surface-copy.

I want to ultimately create a set of macros (or a DSL as the case may be) that will take a description of a process (such as mirroring a surface, doing a gaussian blur, etc) and magically generate code that is not only correct with respect to endianness, pixel format, transparency, bit-depth and such, but also tightly optimized.

Optimization has been a serious problem, but it's interesting to consider that a DSL written in Common Lisp might be able to compile code that runs >90% the speed of the equivalent C, but which is also maybe half as verbose since it wouldn't require specifying different versions for 8, 24, and 32 bit images... or even type declarations, ideally.

This only occurred to me a week ago— I'm not sure it hasn't been done yet. It's also wildly speculative, since I have no experience writing DSLs. Should be fun.

Re: Any New Insights

Posted: Tue Aug 30, 2011 9:44 am
by LearnYouALisp
Duke wrote:I'm currently procrastinating on implementing a lot of stuff that already exists in SDL. I've already done lines, circles, and polygons, then I moved on to reinventing surface-copy.

I want to ultimately create a set of macros (or a DSL as the case may be) that will take a description of a process (such as mirroring a surface, doing a gaussian blur, etc) and magically generate code that is not only correct with respect to endianness, pixel format, transparency, bit-depth and such, but also tightly optimized.
I don't know the specifics involved, and you may not be interested, but there is a request for SDL bindings to Racket:

https://github.com/plt/racket/wiki/Intro-Projects