[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Pnet-developers] Re: Pnet-developers Digest, Vol 24, Issue 2
From: |
Roman I Khimov |
Subject: |
[Pnet-developers] Re: Pnet-developers Digest, Vol 24, Issue 2 |
Date: |
Sun, 5 Nov 2006 21:13:20 +0300 |
User-agent: |
KMail/1.9.5 |
Kirill Kononenko:
> Roman, to strip features from your tuned verson of Portable.NET one
> need to formulate a list of features which you want to have available
> on your device or features which you do not want to support.
That's for sure, and I have some picture in mind for what I want to see as a
final result. But as it's somewhat experimental/educational project, I want
to check first what I have now and what can be removed easily.
So that was the first thought when I looked at verify.c - why should I have
verifier when I don't need it? I just hadn't knew how close it is related
to interpreter itself. :)
> To sum up, if you shave every possible bit of PNET you would just get
> something like a program which returns 0 after a main(). So you need
> to write down if you want, for example, 32-bit, 64-bit, float numbers
> to be available, delegates, threads, internal functions and pinvokes.
> Then after you have this list you can start shaving the source code of
> the engine and working on the libraries. A very special, optimized and
> reliable version of PNET could be done but it needs knowledge, work
> and time.
But at first some bugs need to be fixed, as non-FP compilation (and I don't
need floating point for sure). From what I see currently the main and
easiest thing to be stripped is pnet support libraries, for example there
is just too much crypto stuff. I want to make some library components
optional, configurable items in chance such patches would be considered
useful by pnet developers.
--
Roman
http://roman.khimov.ru .o.
mailto: address@hidden ..o
gpg --recv-keys 0xE5E055C3 --keyserver hkp://subkeys.pgp.net ooo
pgpaTU3OP2PXg.pgp
Description: PGP signature