This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Inserting new parameter in gcc
On 09/11/2007, Basile STARYNKEVITCH <basile@starynkevitch.net> wrote:
> Manuel López-Ibáñez wrote:
> >
> > RejectNegative is not intuitive? I really would like to hear your
> > suggestion. Honestly, not sarcasm here.
>
> I agree with the original poster Juan Luis Liarte, RejectNegative, as a
> marker for string argument, is extremely counterintuitive: strings are
> not negative (nor positive)! So I welcome another keyword form them.
> Perhaps just StringArgument could be ok?
>
> Is there some case where RejectNegative does what it suggest, ie accept
> (only) numerical (integer) arguments and reject them if they are less
> than 0?
Huh? RejectNegative rejects the negative form of a parameter. That is,
-fno-myparameter won't be accepted. Having a negative form is the
default for -f* and -W* parameters. I am not sure if that is
documented but it should be documented and referenced at the beginning
of every *.opt file.
> > Also, you would need to be versed in awk, since the scripts that parse
> > the *.opt files and generate C code for options are written in awk. Or
> > alternatively, you would need to replace them with something else. But
> > that won't be a little patch anymore.
> Even if it is not a tiny patch, I would welcome it (but I am not in
> position of formally approving it).
I cannot be against a patch that hasn't even been described but keep
in mind that tiny patches are: safer, easier to review and faster to
approve.
Documenting the current syntax would be an improvement. If you need
help preparing a patch, just ask questions in this mailing list.
Cheers,
Manuel.