<contingencyplan> Hey, I'm thinking I'll be able to have SeqL -> C generation done by the end of next week. Shortly after that, I'd like to do an IRC-based SequenceL seminar, if you're up for it.
* morgul|laptop misses playing halo
<morgul|laptop> Sure.
<contingencyplan> kickass
<contingencyplan> I'll even see if Brad or Bryan (or maybe even Dr. Cooke) can stop by for a bit.
<morgul|laptop> We should try to get some people from CS channel in here for that.
<contingencyplan> Yeah.
<morgul|laptop> That would be awesome.
<contingencyplan> I'll set up a time and such - realize that the C code I'll be generating will be shit compared to hand-written code, BUT it will get better from there.
<morgul|laptop> I would be interestedin having a few handwritten examples to compare speeds with the generated code.
<contingencyplan> it will not be as fast. I can guarantee that the handwritten will be faster.
<morgul|laptop> I know that.
<morgul|laptop> but I want to see how much faster we're talking.
<contingencyplan> gotcha.
<morgul|laptop> 20%? 100%?
<morgul|laptop> I would also like to compare it to Lua and Python versions of the same thing.
<contingencyplan> hrm... not 100% - I can get the numbers between another interpreter vs. hand-written NASA code
<morgul|laptop> See how it currently falls in terms of speed.
<contingencyplan> hrm. that latter will be a bit harder.
<contingencyplan> not too difficult to do, but my focus will be getting the C code generated first.
<contingencyplan> At any rate, I'm heading out.
<morgul|laptop> you misunderstood
<contingencyplan> oh?
<morgul|laptop> handwritten demo code in python
<contingencyplan> hrm. that's comparing apples to oranges.
<morgul|laptop> and compare to geerated c code from SL
<morgul|laptop> generated*
<contingencyplan> hrm. That's more a benchmark for C vs Python.
<contingencyplan> But when I have SL -> Python (shouldn't be hard after getting C), then we can do comparisons there too.
<contingencyplan> But the main thing is comparing generated to hand-written, regardless of the language.
<morgul|laptop> ok
<morgul|laptop> yeah
<contingencyplan> If we change languages mid-stream, then the comparison is worthless.
<morgul|laptop> yeah
<morgul|laptop> I was thinking more like compart GenC -> HWpython, HWC ->HWpython. It migh prove interesting, if worthless.
<contingencyplan> But once I figure out how to get the optimizations (eliminating common sub-expressions, for instance), then we'll be able to achieve parity (or near-parity) with HW code
<contingencyplan> Well, it might prove interesting, but we can do that later. I don't want to try doing that for the seminar - stick with C only for that.
<contingencyplan> Otherwise it 1) wastes time (since it is worthless overall) and 2) adds more things to confuse people.
<contingencyplan> SequenceL will be enough of a paradigm shift to confuse the uninitiated. I don't want to add to taht.
<contingencyplan> *that
<contingencyplan> At any rate, I will talk to you later - catch you later tonight.
<morgul|laptop> ok