Re: versioning config.cache

[Harlan Stenn]

Eric Blake wrote:
> According to Dave Hart on 5/1/2009 9:15 AM:
> >> Since this variable is under your complete control, I would have written
> >>
> >> if test $ntp_cache_flush = 1; then
> >>  ...
> >> fi
> >>
> >> rather than a case statement.
> > 
> > That's a fine choice.  I think for my purpose I still prefer the case
> > for legibiity, though i realize it's inefficient.
> 
> In the grand scheme of things, 'if test... fi' and 'case ... esac' are
> both shell builtins, so no change in efficiency on the number of processes
> spawned.  ...

The problem is that the grand scheme of things is recent.

In older shells 'test' was often *not* builtin and as these older
shells run on older systems the performance hit was is definitely
noticable.

This topic is something Larry Wall brought to my attention when I was
doing the 2.0 release of metaconfig in preparation for the original
release of perl4.  Just to offer some context.

Anyway, coupling this with the fact that NTP's configure script is on
the huge side of massive, and has *lots* of test cases like this...

Dave Hart can probably tell you how long the configure runs are taking -
I'm kind of used to them so I don't think about it anymore.  But I do
get regular complaints from people about both the size and the runtime
of our configure script.

H