[m-rev.] for tryout and review: don't include some unneeded items in .int3 files

[Peter Wang]

On Mon, 18 Mar 2019 17:04:40 +1100 (AEDT), "Zoltan Somogyi" <zoltan.somogyi at runbox.com> wrote:
> Julien, Peter, I would like to ask you both to try out the attached diff
> on the Mercury programs you work with, and report whether a compiler
> that has this diff can compiler those programs without problems.
> If there are no problems with any of your programs, *then* I would ask
> for a review.

It compiled the programs I work with just fine.

> 
> This is the first installment of stage 2 in the plan I posted a while ago
> to fix our system of interface files. I have a bunch more similar
> changes in mind, for which I would make the same request of you both.
> I could either let each logical change stand on its own a diff, or I could
> create batches of changes. The former would require you guys to try out
> more compiler versions, while the latter requires more detective work
> for me in the event of a failure (to figure out which *part* of the batch
> failed). The latter would also require more compiler tryouts to confirm
> the validity of any fixes for the failures.
> 
> I am leaning towards grouping 3 or 4 changes into batches.
> Does anyone have another preference?

git bisect works better with smaller changes. You can also push a
series of provisional changes to a separate branch without waiting
for confirmation for each change.

Peter