Re: Testcase minimization with Delta

[Daniel Berlin <dberlin at dberlin dot org>]

On Sat, 16 Nov 2002, Zack Weinberg wrote:

> Daniel Berlin <dberlin@dberlin.org> writes:
> 
> > On Sat, 16 Nov 2002, Daniel Berlin wrote:
> >> 
> >> You didn't run topformflat on the program, did you?
> > topformflat, in case you didn't notice, will flatten the program to 
> > whatever level (IE it makes it so line breaks are only allowed at certain 
> > level.  Thus, flattening to level 0 will make each function a single 
> > line.).
> 
> Yes, I used topformflat -- that's how I got it from 32,000 to 5,000
> lines.
Err, then that didn't change the testcase at all.

>  Unfortunately the -level switch refused to do anything, so it
> was not possible to get finer grain that way.
Yes, i know, that's a pain in the ass right now.
>  I suspect it would have
> worked better if it had.

Possibly.
You also complained that it iterates removing one line at a time.
This is true, but only after it's tried 32 lines at a time, etc.
The python version uses a newer version of the algorithm which is smarter 
in the granularity it chooses, but in the end, it still has to make a 
final pass removing single lines a time.
This is because the minimal testcase is going to be 1-minimal, meaning 
that removing any line would change it.

Now, if you don't want the absolute minimal, i can make it stop at some 
higher granularity.
> > zw
>