Re: [PATCH v2 1/2] format-patch: have progress option while generating patches

Jeff King [off-list ref] writes:

If it's smooth, the (50,1) case is slightly nicer in that it puts the
progress in front of the user more quickly. I'm not sure if that's
actually worth pushing an additional decision onto the person writing
the calling code, though (especially when we are just now puzzling out
the method for making such a decision from first principles).

So I'd vote to drop that parameter entirely. And if 1 second seems
noticeably snappier, then we should probably just move everything to a 1
second delay (I don't have a strong feeling either way).

Sounds like a good idea to me.  

I've already locally tweaked Kevin's patch to use (0,2) instead of
(0,1) without introducing the simpler wrapper.  It should be trivial
to do a wrapper to catch and migrate all the (0,2) users to a
start_delayed_progress() that takes neither percentage or time with
mechanical replacement.

On Mon, Aug 14, 2017 at 03:42:14PM -0700, Junio C Hamano wrote:

Jeff King [off-list ref] writes:

quoted

If it's smooth, the (50,1) case is slightly nicer in that it puts the
progress in front of the user more quickly. I'm not sure if that's
actually worth pushing an additional decision onto the person writing
the calling code, though (especially when we are just now puzzling out
the method for making such a decision from first principles).

So I'd vote to drop that parameter entirely. And if 1 second seems
noticeably snappier, then we should probably just move everything to a 1
second delay (I don't have a strong feeling either way).

Sounds like a good idea to me.  

I've already locally tweaked Kevin's patch to use (0,2) instead of
(0,1) without introducing the simpler wrapper.  It should be trivial
to do a wrapper to catch and migrate all the (0,2) users to a
start_delayed_progress() that takes neither percentage or time with
mechanical replacement.

I was actually proposing to move (50,1) cases to the simpler wrapper,
too. IOW, drop the delayed_percent_treshold code entirely.

-Peff