Hi folks, I was just thinking that it would be nice if a we had a simple naming rule for 16 bit functions: "A function is 16 bit if and only if it's name ends in 16." This way, we can have more reliable tools for detecting 32-16 bit cross calls (unfortunately, winapi_check fails in many instances for this check). Most 16 bit functions already adhere to this naming convention, but there are still a few that don't. Alexandre, would renaming such functions be acceptable? If so, I can add yet another janitorial task... -- Dimi.
"Dimitrie O. Paun" <dpaun(a)rogers.com> writes:
Most 16 bit functions already adhere to this naming convention, but there are still a few that don't. Alexandre, would renaming such functions be acceptable? If so, I can add yet another janitorial task...
I don't know, I guess that depends how many need to be changed. Do you have a list? In any case, note that we don't have to rely on winapi_check for that; 16-bit functions should be split to separate files so that we can have --disable-win16 do the right thing, and then cross calls will be found by the linker. -- Alexandre Julliard julliard(a)winehq.com
On March 14, 2003 11:26 am, Alexandre Julliard wrote:
I don't know, I guess that depends how many need to be changed. Do you have a list?
No, I'm afraid I don't. In particular, I noticed that the functions in loader/ne/*.c don't have the 16 prefix, even if most/all of them are 16 bit functions, AFAICS.
16-bit functions should be split to separate files so that we can have --disable-win16 do the right thing, and then cross calls will be found by the linker.
Right -- but how do we make sure we've split all 16-bit functions to their own files? Once they are in the 16-bit only files (do we have a consistent naming convention for those, so we know they are 16-bit?), there's not that much benefit in renaming them, I agree. I guess what I'm saying is that if we had these rules: Function names end in 16 IFF they are 16bit functions File names end in 16 IFF they get compiled out by --disable-win16 It makes it so much easier to spot problems by just scanning the code, not to speak of the simplicity that it brings to any analysis tool. -- Dimi.
"Dimitrie O. Paun" <dpaun(a)rogers.com> writes:
No, I'm afraid I don't. In particular, I noticed that the functions in loader/ne/*.c don't have the 16 prefix, even if most/all of them are 16 bit functions, AFAICS.
Ah, so you want to change internal functions too? I don't think I agree with that.
I guess what I'm saying is that if we had these rules: Function names end in 16 IFF they are 16bit functions File names end in 16 IFF they get compiled out by --disable-win16 It makes it so much easier to spot problems by just scanning the code, not to speak of the simplicity that it brings to any analysis tool.
I don't think uglifying the function names just to make the tool easier is a good trade-off. There are ways of doing the analysis by using the linker, and that's much better than an approach based on function names. -- Alexandre Julliard julliard(a)winehq.com
participants (2)
-
Alexandre Julliard -
Dimitrie O. Paun