Nonrecursive build within gcc directory

Zack Weinberg zackw@Stanford.EDU
Sun Oct 22 15:34:00 GMT 2000


On Mon, Oct 23, 2000 at 04:34:29AM +1100, Tim Josling wrote:
> This is great. It will save having to put a lot of redundant
> information, twice, in two places.
> 
> But is it really not possible to optionally turn off pedantic and
> traditional for the front ends.

I'm afraid not, unless you want to force everyone to use an extremely
recent version of GNU make.

> Before I turned them off for my nascent COBOL front end, I was
> getting thousands of entirely useless messages. I am not planning to
> support a K&R compiler - after all ANSI C has been around since the
> decade before last - and I do not want to limit myself to features
> supported in K&R.

I think you'll find that you can avoid all the warnings without having
to jump through too many hoops.  The C front end and the optimizer are
all written to avoid most or all those warnings, and we still manage
to use prototypes, etc.

> As for pedantic I have never found any of these messages helpful
> (others may I suppose).

Examples?

> A lot of the messages also relate to the output for flex and
> bison and other utilities which I cannot control. 

*boggle* flex/bison always generate warning-free code for me...

> By the way can anyone tell me how to get rid of the message
> 
> 'int constants are signed in ansi unsigned in K&R' 
> 
> (or maybe its the other way around)?

Write unambiguous integer constants.  This is always possible.

> It seems to report the error even when defining a constant that has
> the same effect regardless of dialect

It shouldn't.  Examples?

zw


More information about the Gcc-patches mailing list