minimal requirements for build C compiler?

Rob Landley rob at landley.net
Fri May 26 15:34:24 UTC 2006


On Thursday 25 May 2006 9:31 pm, Mike Frysinger wrote:
> On Friday 19 May 2006 17:08, Rob Landley wrote:
> > On Thursday 18 May 2006 9:39 am, Robert P. J. Day wrote:
> > >   what C standard is the build compiler for busybox expected to adhere
> > > to?  more specifically, can i expect C99 so i can use designated
> > > initializers for aggregate types?
> >
> > Yup.  We went to c99 in 1.1.1, when we dropped support for gcc 2.95.
>
> not to dig up the previous thread, but you're willing to drop support for
> software released in 2001 (gcc-2.95.3) but not software released in 2000
> (make-3.79.1) ?


A) What I'm not willing to drop support for yet is Red Hat 9 as a build 
environment, which according to the dates on the iso images at 
archive.download.redhat.com was released in September 2003.

B) c99 is a spec that's been official for seven years, and thus dates to 
before any of the actual implementations in question.  Nobody's proposing 
going to a new _spec_ for make, they're proposing requiring a specific 
version of one implementation of it.  The move to c99 allows us to be _less_ 
tied to gcc, because there's a spec.  We dropped 2.95 because it didn't 
support that specification.

C) I'm a bit uncomfortable about requiring such a narrow range of versions of 
a single tool (GNU make, either the most recent two dot releases or a CVS 
snapshot).  Gcc has 3.2.x, 3.3.x, 3.4.x, 4.0.x, and 4.1.x each with as many 
or more dot releases than make 3.8.x, plus various clones of its syntax 
(Intel's C Compiler, TCC) that I'd be interested to hear the results of 
people trying to build under.

Does that help you understand my position?

> -mike

Rob
-- 
Never bet against the cheap plastic solution.



More information about the busybox mailing list