Re: [tclquadcode] Defaults cause confusion on dkf-callframe-impl branch

[Donal K. F. <don...@ma...>]

On 29/05/2017 04:51, Kevin Kenny wrote:
> (2) 'mrtest::*' is hanging in the specialization phase for some reason.
> I suspect that there's still some line of code lurking about that's
> failing to account for the callframe arg on 'invoke'. I'll track this
> one down.

A teeny bit of trivial instrumentation in computeTypes says that the 
while loop is going back and forth between these work packets:

    0 ::mrtest::calc -1879244801
    0 ::mrtest::fibo -1879244801

Now, it's trivial to detect that this is what is going wrong, but I'm 
not sure what to do with the detection or why the failure might be 
occurring. The simple-minded instrumentation I added indicates that 
we're currently analysing many procedures far more than I'd expect; over 
a hundred times in the case of “::linesearch::getAllLines1 -1879244801” 
(and a little less with [getAllLines2]). I'm guessing this has to do 
with the calling of [::linesearch::collinear] from within several nested 
loops, but I'm not clear why this would be so.

I've put the instrumentation I was using on the 
dkf-specializer-instrumentation branch, but that's mostly so you can see 
what I was looking at; I didn't write it to be anything other than 
intrusive and noisy. ;-)

Donal.