Re: [Gcl-devel] Re: sequences / in-lining

[Paul F. Dietz]

Camm Maguire wrote:

> Also inlined are: length and reverse, endp and identity are expanded,
> the former to throw an error only when safety >=1.
>
> There is an unused heap sort in the code too, but I can't see any gain
> here, as the memory requirements are virtually the same.  The
> quick-sort is non-recursive and uses a tiny stack ~ log_2(n).  I have
> to read up on what stable-sort is supposed to be.
>
> More to follow.  Feedback appreciated.  Notably, sort is likely the
> biggest inline we would ever consider, and its impact on code size
> might be arguable.  But it certainly pays off in performance.

Build time for gcl on my machine seems to have gone up by roughly
50%.  The size of the .o file for the random tester appears
to have doubled.  Is the inlining being a bit too aggressive?

        Paul