memory functions: smax=0 should be valid

[rurban]

memory functions: smax=0 should be valid

[rurban]

memset_s documents the different behavior if under C11:
Since C11 smax=0 is allowed, before it fails with ESZEROL.
So we'll have to check for C11 (added a compile-time configure probe)

On C11 allow smax/n=0, before keep the old behavior: ESZEROL. Good?

[rurban]

I don't care that much about the perf. issue, but about convenience and conformance, analog to free(NULL) (supporting your point).

I'll check how prior art (MSVC) does it, when the standard does not mention it.
And this code was in production at Cisco since 2008, so changing might be disruptive.

Using memcpy_s over memcpy might suggest erroring on a 0 length argument. i.e. a security feature. (keeping it asis)

[rurban]

In MSVC and wine/ReactOS smax=0 is valid: https://msdn.microsoft.com/en-us/library/wes2t00f.aspx

[jeremyhannon]

fixed in d453998.

[jeremyhannon]

@rurban since you now have a HAVE_C11 and WANT_C11 I was a little confused about the difference until I researched it. So I'd suggest renaming WANT_C11 to WANT_C11_ANNEXK or WANT_C11_LIB_EXT1 to make it clearer that you're wanting a subset/extension instead of wanting to use the C11 base standard.

But at that point why not just use #if __STDC_WANT_LIB_EXT1__ >= 1 everywhere so you're not using a custom symbol?

[rurban]

I admit it's confusing at the first glance.
But I'd rather stay with the short form. It's only used in a few places where the actual upstream code is usable (memset_s on darwin).

WANT_C11 means #if __STDC_WANT_LIB_EXT1__ >= 1, yes. But we would need to use the long form #if defined(__STDC_WANT_LIB_EXT1__) && (__STDC_WANT_LIB_EXT1__ >= 1)

The idea of safec is to implement C11 fully. Other's say they do, but they don't. Having this discrepancy, I think it's easier to make it clear with this short name.
I don't want to emphasis it as extension, safe functions must be the default. Microsoft even warns all users to switch over to the safe variants.