Re: [Automake-NG] [PATCH 00/11] Several changes to parallel-tests support

[Bob Friesenhahn]

On Thu, 10 May 2012, Stefano Lattarini wrote:
> > > These numbers are still acceptable IMHO, and are a price worth to paying
> > > for the improved cleanliness of the code.
> >
> > 10 seconds would not be at all acceptable to me.
> But notice it is 10 seconds for *12 runs* and with *5 thousands tests*.  That
> is already a very corner case situation, not a typical one.

I did not notice any description of what the measurement of these make 
targets entailed.

The main metric which is of interest to me is how long 'make all' 
takes after everything has already been built and there is nothing to 
do.  Time to rebuild one source file and re-link adds directly to 
that.  Perhaps I am atypical due to being a developer, but this is 
something that I sometimes do hundreds of times in one day.

As an analogy, this is simiar to how long it takes between when you 
press on a car's accellerator and when it starts to actually speed up. 
If it took 2 seconds, you would start to feel unhappy.  If it took 10 
seconds you would be quite unsatisfied (and perhaps soon, dead).

> In my mail I drafted such a plan (at least for the kind of issues under
> discussion here, i.e., deeply nested lazy-eval variables evaluated several
> times).  Didn't that sounds reasonable to you?  If not, why?  Honest
> question, not rethorical.

I have not yet taken time to read and understand the plan.

Bob
--
Bob Friesenhahn
address@hidden, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/