Re: Using .opt files for target options

[Neil Booth <neil at daikokuya dot co dot uk>]

Richard Sandiford wrote:-

> The idea is that the new machinery will work alongside TARGET_SWITCHES
> and TARGET_OPTIONS for the time being.  We can remove things like the
> "" hack if (when?) we remove support for TARGET_SWITCHES and
> TARGET_OPTIONS.
> 
> The alternative would be to remove all "" entries from TARGET_SWITCHES
> on-masse, but I'd much rather move each port individually.
> 
> > I don't see why you needed to change this.  I believe the existing
> > code only does this for -f and -W flags, so -m should be unaffected.
> > Can you explain?
> 
> I forgot to mention this in the covering note, but the patch extends the
> same behaviour to -m.  Most -m flags have the same kind of negatives as
> the -f flags; certainly most of the MIPS ones do.
> 
> The point I was trying to make was that the automatic negative of
> "-mips16" would be "-mno-ips16", not "-mno-mips16", so we need the
> flexibility to list unusually-spelt "-mno-" options separately.
> Normal -f, -m and -W negatives will still work OK.
> 
> FWIW, I don't think the overhead of two look-ups is going to be
> significant.  It also doesn't need much extra code (+3 lines, -1 line).
> 
> > Is there a good reason you don't simply enforce it being called
> > target.opt?  Wouldn't that be simpler?
> 
> The idea is that you'd put options that apply to all i386 targets
> in i386.opt and those that apply only to cygwin (such as -mcygwin)
> in cygwin.opt.

Thanks for the clarifications.  This is OK.  Please do try to transition
as much as you can stomach.

Neil.