Re: demand_empty_rest_of_line and ignore_rest_of_line

[Hans-Peter Nilsson <hp at bitrange dot com>]

On Sat, 24 Apr 2004, Zack Weinberg wrote:
> Hans-Peter Nilsson <hp@bitrange.com> writes:
> > See binutils archives from last time this came up (search for
> > "no_app").  It saves as much as 1% off the compile time.  I
> > think *more* targets should use it, but most would need to tweak
> > their md:s to avoid redundant spaces.
>
> Frankly, (as someone who has to stare at GCC's assembly output all the
> damn time), I do not think 1% speedup is worth the *severe*
> degradation in readability that this would impose.  I would much
> rather put effort into making GAS's parser be faster in
> input-scrubbing mode.

I think you overrate the amount of change needed.  As another
person who has to stare at GCC's assembly output all the (...)
time, my take is that it would (if anything) improve readability
;-) but most problably it's neutral maintainerwise.

All it takes is to change some spacing like "move.d r0, r1" to
move.d r0,r1" and remove spurious (often just extraneous)
spacing in target-defined pseudoopcodes and rarely used
constructs like the trampoline template.

Consider checking cris.md and pointing out what #NO_APP-related
spacing makes it "severely" unreadable!

Not that speeding up GAS parsing would be a bad thing, of
course.

brgds, H-P