Re: [cc65] Alignment

Date view Thread view Subject view

From: groepaz (groepaz_at_gmx.net)
Date: 2002-07-18 21:36:05


Hello Ullrich,

Thursday, July 18, 2002, 9:19:45 PM, you wrote:

UvB> other compiler packages need when doing alignment (groepaz is right when he
UvB> says that gas can do alignment, he forgot to tell that segments are aligned at
UvB> 16 byte addresses for DOS and at full memory pages for most other operating
UvB> systems, so gas has the needed prerequisites).

well, so what? align sections to 2 byte boundaries (kills indirect
jump problem for eg) then for a start (we can live with those 3 bytes
wasted in typical c program then ;=))

other than that, i am using gcc with the gba, the gp32, the psx etc
blabla and i can very well place binaries whereever i want and
alignment works just fine (i am not saying that the linker wouldnt
simply always align to a certain common denominator, if thats the only
way it works, so be it)

UvB> If someone thinks that ca65 is doing something wrong, please try first to
UvB> understand how alignment is currently implemented and then tell me why you
UvB> cannot live with that. Until someone comes up with a real good reason why the
UvB> current implementation is unusable, I would prefer not to waste any more time
UvB> on this discussion.

certain stuff, such as routines that are heavily optimized for speed -
or even the indirect jump thingy - just _need_ properly working
alignment. as simple as that :o) as its now, any of these routines are
useless in a typical c-program, just because it loads to an odd
adress. thats just plain whacked :o]

-- 
Best regards,
 groepaz                            mailto:groepaz_at_gmx.net


----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo_at_musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.


Date view Thread view Subject view

This archive was generated by hypermail 2.1.3 : 2002-07-18 21:36:18 CEST