On May 16, 2011, at 2:53 PM, Jeff Morriss wrote:
> But it is encouraging as they indicate that Solaris is the only "significant" printf that is "good enough" except for this NULL pointer stuff. Well, except for those poor sods running "insignificant" printfs.
To be fair, what Owen said was "Solaris is *probably* the only significant example of a printf that is good enough otherwise but doesn't catch null." (emphasis mine) Given that the bug was for Win32, he probably meant "the only significant example of a UN*X printf"; the UN*X printfs I know of do:
glibc - (null)
*BSD libc (including OS X and iOS) - (null)
Solaris - crash
My guess is that at least some of the other commercial UN*Xes also crash, as I think substituting (null) was either a 4.4-Lite-vintage BSDism or a glibcism. At this point, "the other commercial UN*Xes" probably means "AIX and HP-UX" (well, and the SCO UN*Xes, but I suspect the free software community is rather uninterested in them, given the whole SCO vs. Linux noise); I don't think any others are actively being developed. I don't know whether either of those are considered "good enough otherwise", which I assume means "we don't just replace it with ours".