On Saturday 27 December 2008, Tamas TEVESZ wrote: > On Sat, 27 Dec 2008, Dan Pascu wrote: > > > wfontpanel.c:588: warning: missing sentinel in function call > > > > This is a very interesting warning, considering that NULL is defined > > like: > > not very much interesting, but you are partially right, it should have > been (char *)NULL instead. gcc (since at least 3.4 on linux) already > defines NULL as (cahr*)0, that's why i assumed NULL would be ok, but > that doesn't guarantee anything. > > 0 (whether as 0 or NULL defined to 0) is definitely wrong, as on 64bit > archs (or, more precisely, lp64 archs) the size of an integer is not > the same as the size of a pointer.
You are right. Using 0 will give a warning on 64 bit architectures. NULL won't however as it it really defined as: #define NULL (void*)0 > > > #define NULL 0 This definition is only used when __cplusplus is defined. > > > > On what OS do you get this and with what compiler? > > this particular setup is x86_64 linux with gcc 4.3.2, but (especially > considering the above) i'm pretty sure that if and when i manage to > put together my sparc, alpha and mips gear (all of them tend to expose > different "troublesome" areas towards stuff that have never really > considered anything other than lp32 little-endian non-strict alignment > archs), they will stir wm pretty well. > > obviously it would be helpful to have someone with more portability > experience than me (anyone care to recruit one?). > > dan, i think i can speak for all of us when i say that we respect your > decision of not being involved anymore, and we are grateful for the > occasional comments and such. with that in mind i only have two > questions left, which if you answered, would be of great help. > > (as far as i can see now) a great deal of confusion comes from the way > the width and length of the images (RImage) is handled all over the > place. so giving a hint whether they are ints or longs (question #1), > signed or unsigned (question #2). having an answer to these would be > much help. I wouldn't bother trying to fix these. Before we handed over the project (Alfredo and me) started work to replace wrlib with cairo. This is the path I suggest you follow, as it's not worth implementing and maintaining a raster library when one can use a more complete and better optimized solution like cairo. Our (work in progress) is available here (as a darcs repository): http://devel.windowmaker.info/cgi-bin/darcsweb.cgi?r=wm2;a=summary -- Dan -- To unsubscribe, send mail to [email protected].
