Re: bison-generated code causes compilation failure on OpenBSD 4.7

[Eric Blake]

On 10/07/2010 03:19 PM, Jim Meyering wrote:
> > Stepping back, malloc shouldn't even be declared here, since
> > this skeleton code has already included stdlib.h.

Agreed.  This portion of bison output stems back to K&R days, when you 
couldn't guarantee things like malloc's declaration.

But given that more than a year ago, Bison was debating about switching 
to generating code that assumes C89 or better[1], this is yet another 
argument in favor of moving in that direction - by completely dropping 
the malloc() declaration and instead relying on <stdlib.h> to do it's 
job, this problem will be trivially avoided.

[1] http://lists.gnu.org/archive/html/bison-patches/2009-08/msg00085.html

--
Eric Blake   address@hidden    +1-801-349-2682
Libvirt virtualization library http://libvirt.org