Larry Wall <larry@wall.org> writes: >Nick Ing-Simmons writes: >: This was presumably to avoid the extra >: test so that GvSV was quicker than GvAVn - but this is of less value >: now that "most" scalar vars are presumed to be lexical. > >Yes, I want lexicals faster than globals, but you guys keep coming up >with ways of slowing down globals instead of speeding up lexicals. :-) I am speeding up lexicals by not filling the cache lines with a bunch of SVs which are never referenced ;-) Attached is a toy module. Used like this : perl -MGVTax -e '' 48 GVs of which 8 (16.6667%) are defined or a partly exercised Tk demo: perl -MGVTax basic_demo ... 2807 GVs of which 110 (3.91877%) are defined I am not about to argue that 3000 SVs is going to amount to much, so if that is whole scope of the problem I will shut up again. > -- Nick Ing-Simmons <nik@tiuk.ti.com> Via, but not speaking for: Texas Instruments Ltd.Thread Previous