Re: writev/readv addition to file_operation proposal

[James Manning]

[ Tuesday, February  8, 2000 ] Abramo Bagnara wrote:
> Without any official answer, with no objections and the enthusiastic
> welcome of LWN, I'm a bit embarassed about what to do now.
> 
> I'm unable to understand if a missing answer need to be interpreted as a
> silent refusal, as an invitation to write down the patch or as a kick in
> the ass ;-)
> 
> Or perhaps only an invitation to wait some time...
> 
> Are there some silently refused proposal veteran that offer me a
> shoulder to cry over or some consolation words? ;-)
> 
> Linus, don't show me a rock heart... how is it possible to not answer to
> a crying man?

FWIW, based on these additions to file_operations I had made a patch
to make all file_operations initializations into named initializers
(2 patches, one gcc and one c99) so that you could add the readv/writev
wherever you wanted, not forced to append past lock (which I hate since
so many of these struct types are now well past a single cache line)

If it makes you feel any better, it's been equally ignored :)  I've
made 3 versions, against .41-pre2, .41, and .42

Perl script:
	http://sublogic.com/patches/c99_pl.txt
gcc syntax patch:
	http://sublogic.com/patches/file_operations-2.3.42-gcc.patch
	 167 files changed, 980 insertions, 2416 deletions
c99 syntax patch:
	http://sublogic.com/patches/file_operations-2.3.42-c99.patch
	 182 files changed, 1061 insertions, 2497 deletions
(more since it takes a little over a dozen instances from gcc -> c99)

James