This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Revised patch for IA64 -mb-step option


On Tue, 2004-04-13 at 17:54, Zack Weinberg wrote:
> not helping anything.  The new behavior is to silently ignore all
> unrecognized -m options, and complain about all other unrecognized
> options,

I looked at all gas target ports, and I see none that do this.  I
understand why you want to do this, to avoid gcc/gas synchronization
problems, but this does not appear to be the right solution.  It appears
to violate established gas conventions.  It also may result in silent
problems with gcc/gas option passing if gas silently ignores options it
doesn't understand.  The existing convention seems to be to require
people to update gas in sync with gcc if they want to use new options. 
This can be inconvenient at times, but does seem to be workable.

> rather than silently exiting unsuccessfully.  (N.B. the code
> in gas/as.c handling an 0 return from md_parse_option appears to be
> broken, but I'm not sure what the right fix is.  Unless the
> expectation is that md_parse_option will have issued a diagnostic if
> it returns 0, which is not what tc-ia64.c was doing.)

Having just looked at all gas ports, I would say the convention is that
md_parse_option must call as_bad (or as_warn) for unrecognized options
before returning 0.

For IA-64, it is simpler to just take m out of md_shortops, as you have
done all of the work of moving everything else to md_longopts.
-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]