This is the mail archive of the
mailing list for the GCC project.
David Edelsohn wrote:
>>>>>> Alexandre Oliva writes:
>>> Except that no_new_pseudos was not used consistently.
> Alex> I'm not sure what you mean by "consistently", but regardless, how
> Alex> could any argument possibly make it better to replace it with
> Alex> (reload_in_progress || reload_completed)
> Alex> rather than
> Alex> !BEFORE_RELOAD_P()
> Alex> such that anyone willing to clean things up from this point on may
> Alex> have to use complex grep patters to account for ongoing line breaks,
> Alex> reformatting and any other boolean transformations, just because it
> Alex> had to be expanded right away rather than retained as a macro?
> !BEFORE_RELOAD_P() is a conservatively correct replacement for
> no_new_pseudos, but not the best option. Each location will have to be
> audited based on context, so your proposed change does not help as much as
> you imply.
I agree, i am objecting to the syntax of what you are saying. I am
objecting to your assertion that it is any real improvement.
The work here is not changing the bits. the work here is the actual
auditing of each place to see if it was the correct place.
I did the auditing for the middle end and did not do it for the backend,
except where i blew it in the alpha.