2009/7/30 Slava Pestov <[email protected]>:
> Hi Paul,
>
> Tests were failing on Windows because some compiler changes I was
> working on resulted in an image size increase. I wanted to postpone
> looking at this until the compiler was done, but this took longer than
> I hoped. Indeed, with my recent improvements the size went down again
> and the tests passed, but then I broke something else ;-) Later today,
> there should be a clean build and binaries will upload.

Thanks! I appreciate that this is fiddly stuff (if it wasn't, I'd be
doing my own build!) so I'm grateful for you taking the time to look
at it.

> Bear in mind that there are still some regressions. I just got inline
> GC checking working again, but I need to do some more testing on this
> first. Also, global stack analysis inserts partially redundant
> instructions in some cases, resulting in a performance decrease on
> some code. I know how to fix this and hopefully I'll do this within a
> day or two.

No problem. I'll keep that in mind. I'm just playing with a problem a
friend suggested to me[1], which is highly CPU intensive. I'm trying
various languages I know, to see how they compare in readability vs
performance terms. So the new optimisations could be very useful (and
if I hit regressions, I'll let you know :-))

Paul.

[1] Simulate dealing and scoring a cribbage hand, many millions of
times. Gather stats on the distributions of the results. Bonus points
for making the hand generation and scoring procedures easy for the
user to fiddle with (that bit seriously hurts languages with high
function call overheads!)

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Factor-talk mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/factor-talk

Reply via email to