[Mono-devel-list] SSAPRE: this time it's for real ;-)

[Massimiliano Mantione]

On Fri, 2004-11-26 at 14:17, Willibald Krenn wrote:
> Well, I already read the summit notes multiple times - it just talks 
> about Paolo doing new trampolines and Zoltan doing a new object file 
> format (any specification/idea available?).

I'm pretty sure (even without checking) that when it mentions the
trampolines work it says it can pave the way to the use of on line
profiling to recompile hot methods.
Also, following this list, I remember Paolo posted a hack of his
own, a statistical profiler... guess what it could be turned into?

> The problem is: If I go ahead and do something in that area and propose 
> a patch on the list, it won't be included because the JIT team has come 
> up with a completely distinct solution to the problem. :-((

I don't see the problem, because... ***

> Everything happens behind closed curtains - you are the only one who 
> actually gives feedback and enjoys a little conversation (with 'nuts' 
> like me) from time to time. Thanks a lot for that!

Hey man, no closed curtains here, really :-)
The simple *existence* of the mono summit document in public
is just the opposite of a closed curtain!

> Ok, so it won't be of any help if I would go ahead implement some algo 
> based on the papers above, I guess. Anyway, I'm looking forward to your 
> post about the idea :-)

*** ...one just cannot write down *everything* he thinks about!
I have *tons* of ideas on things that could be done, and if you'll
implement some of them before I do, I'll just be happy for it!

If I "waste" half a day implementing an algorithm to have a post
dominator tree because I feel I need it, I'm not going to also
spend two more days to fully document all my fantasies on how that
algorithm could be used in conjunction with other ideas (writing
technical things in a clear way takes time).

But rest assured that if I'm going to spend months to do something
concrete, then some info will be posted, as it has been for SSAPRE,
and as it will be for my next SSA work.

So if you follow the list (and the project in general) there's no
risk of doing "useless work", and even less if you also post your
ideas.

What is unfortunate (IMHO) is that we're working on two forked
projects, and *this* can lead to a waste of time and resources.
Anyway, IMHO.

Really, I can't understand your impression of "closed curtains",
from the inside I just don't see them.
On the list (and/or on Monologue and on the project site) I see
nice descriptions of every ongoing activity.
What could one want more?

Ciao,
  Massi