At 01:02 AM 8/2/2006 +0400, Arkady V.Belousov wrote:

>  Though, I don't understand, why to AVB> duplicate all pops on each exit 
> branch, whereas you may just move push      "move _pushf_" AVB> after 
> other pushes?

Because it was not prone to any introduced error that way.  Many pushes in 
many routines occur following, or intermixed with, code which introduces 
side-effects.  I don't seem to be able to impress upon you how important it 
is at this stage to ensure that introduced code has the minimum possibility 
of unwanted side-effects.

You look at a single routine and say "yeah, I could safely move around the 
pushes", but that's not how it works.  I don't look at a single 
routine.  Before going in to correct a code condition at this stage of 
FreeDOS, I look at what is involved in code which might be affected and 
decide, "this is how I'm going to address CY POPF-related routine changes 
right now so things can't break no matter what the push order".  Big deal, 
it costs a few extra bytes in extended memory until the Arkady-influenced 
EMM386 version 3.Oh.My.Goodness comes out AFTER FreeDOS 1.0 release.

If you prefer, think of it this way:  This stuff isn't 
compiler-optimizations like you submitted for Open Watcom where the 
optimizations affect every program everyone builds thereafter.  This is a 
few bytes outside of DOS's memory (you remember DOS, that's what this is 
for) where it can hardly matter.

>  Currently EMM386 isn't complete - there are no working build batch 
> files, so only you yourself may compile EMM386 (and check result).

EMM386 is sufficiently complete for now: it works for the largest number of 
endusers that we can make it work for, which is FreeDOS 1.0's goal as I 
understand it.  The batch and make files aren't for me.  They're for Tom's 
setup and I have my own.  If someone cannot figure out the trivial changes 
needed to rebuild EMM386 properly on their system, they probably shouldn't 
be goofing around in pre-1.0 release source code for a low-level memory 
manager.

I guarantee you that whatever batch and makefile changes you made, and 
however well you think they work, they don't work for everyone.  Open 
Watcom is still iterating through makefile modifications after how many years?

You wish me to make a bunch of last-minute cosmetic and nonoperational or 
nonalgorithmic changes a week or two before FreeDOS has the biggest moment 
in its history?  No, thank you.  Three months back, maybe.  Now, no way.

Those changes are post-1.0 release changes.  I can't be any more clear than 
that.  We'll branch an entire version off for you later if you want, merge 
it back in when it's done and tested.  Go nuts.  Optimize code routines 
down to fractional bytes, cross-platform makefile build EMM386 on an XBox 
360, allow Core Duo multithreading with EFI support if you want.  Just not now.


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Freedos-devel mailing list
Freedos-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-devel

Reply via email to