Looking at the PROC macro in FASM, and some disassembly of C procedures, i see that they use leave to tidy up. However some other compilers don't.

I'm a relative newbie in this area and was wondering why some compilers/coders use leave rather than mov esp, ebp pop ebp

Also, if they use leave why do they not use enter?

Its OK, I managed to track it down after many hours of googlin' at microsoft of all places!

For anyone interested I found it here.

http://www.microsoft.com/msj/0298/hood0298.aspx

Apparently its because leave is smaller compared to its equivilent, but enter actually is slower than the equivilent code.

I still don't get why some HLLs don't use leave, but thats another story i guess.

I think it's a nice article for anyone who wants to learn assembly.

I found this on my travels too, its also on microsoft(strange!)

A blog by someone at microsoft, who seems like he knows what he's talking about. Some of the entrys on windows hacks are very interesting.

http://blogs.msdn.com/oldnewthing/archive/2004/01/08/48616.aspx

Contrary to popular belief , Microsoft actually does have some very skilled programmers. The kernel team are a bunch of pretty decent programmers. Based on his blog, Raymond Chen (OldNewThing) is one of the more skilled guys there as well.

What ruins it, of course, is the incoherent win32 API, and all the bloat ontop of that.

Jay!!1one!

it would be fine if someone wrote a goof procedure entry and exit tutorual. I wrote one, but just for classmates, in Slovak language, and there were not anything about local variables.

I didn't mean to rag on microsoft so much, i'm sure they have many good programmers there, and i really think XP in general is very good. My main issue is with the documentation, which often is very patchy, and as shown many times on Raymonds blog, points developers in the wrong direction (causing the need for the many shims he talks about)

I also found it quite refreshing that Raymond is not squeemish about talking about the bad things, and why they happen.

Raymond's pretty cool, and I also like that he's not afraid to call a hack a hack

IMHO MS did wrong in making shims to make broken software work - they should have let it break, but contact the developers about it. After all, MS are in a decent position to say "hey, we've noticed you do this wrong, you really ought to fix it".