Paul Moore <Paul.Moore@uk.origin-it.com> writes: >From: M.J.T. Guy [mailto:mjtg@cus.cam.ac.uk] >> >> Good point. I was proposing to look for both "PL_na" and "na" etc. >> If a PL... symbol was found, we can assume the module has been updated >> so no POLLUTE is needed. I'd only force POLLUTE if there were >> polluting symbols present and *no* PL... ones. > >But... In the case of na, a good fix for (silly example) > > { na = 12; } > >is > > { STRLEN na; na = 12; } > >instead of > > { PL_na = 12; } > >because access to a local variable is faster than access to a global. > >So a fixed module may still refer to na rather than PL_na. Indeed for na that is what Tk does (mostly - in one or two places I changed the name as well). >(I assume you >don't propose to scan for declarations...?) > >I'm not sure of the value of the heuristic approach in any case, but I >thought I'd add this datapoint... > >Paul. -- Nick Ing-Simmons <nik@tiuk.ti.com> Via, but not speaking for: Texas Instruments Ltd.