From: Keith K. <ke...@ka...> - 2008-05-31 08:19:22
|
H. Peter Anvin wrote: > Andreas Ericsson wrote: > >> I would like to step up and say "I'll do that", but unfortunately I can't, >> as my asm-fu is... well, I don't have any. >> >> I could probably do some janitorial work on the C-sources or some such >> though, although you seem primarily focused on adding new features, so >> I don't know what stuff you would like done, if any, that I can do, >> especially since I'm not at all familiar with the nasm sourcetree. >> >> > > Well, these are things that I think we need: > > - A Mach-O backend maintainer (Mac person). > This strokes my curiosity. I've recently bought an iMac (Intel Core2Duo w/ OS X Leopard) for a pure development machine. I actually have most of the HD dedicated to Windows XP x64 (along with Linux/BSD via VirtualPC) and a healthy amount of space dedicated to OS X for Mac software development. I am currently researching system level programming for OS X in order to include examples in NASMX. The simple part is supporting the BSD-like SYSCALL and mixing that with the X11 examples... nearly killing two birds with one stone. However, I'd also like to include some examples of Mach IPC, Quartz, and perhaps Carbon... while it is still around. At any rate, 64-bit Mach-O support might be a good re-entry point for me in light of my virtual absence from this project. I'll let you know ahead of time, that the entire month of June is booked solid for me in terms of programming. I will keep researching throughout June and approach 64-bit Mach-O in July if no one else has taken it up by then. > - Someone to take responsibility for the instruction table, since we > have a number of issues with it. This one actually requires > relatively little understanding of the code base, especially with the > changes that I've made recently. It mostly requires being willing to > dig through manuals and websites and verify that the table matches > reality. > - Automatic testing. Victor has written a regression test framework, > but we really could use a pile of regression tests (automatically > generating them should be at least partially possible.) > - Documentation. > > There are a lot of core changes which are necessary, such as fixing the > matching system. Right now, a big blocker to those is that some of > those will require instruction table changes, and I'm sorry, I can't do > both. I'm already spending *way* too much time and energy on this. > > -hpa > |