Complete.Org:
Mailing Lists:
Archives:
freeciv-dev:
September 2001: [Freeciv-Dev] Re: Unused field in move_cost_map removed. (PR#965) |
[Freeciv-Dev] Re: Unused field in move_cost_map removed. (PR#965)[Date Prev][Date Next][Thread Prev][Thread Next][Date Index] [Thread Index]
On Tue, Sep 18, 2001 at 12:39:32PM +0100, Gregory Berkolaiko wrote: > I did some experimenting with the returned field of move_cost_map (aka > warmap), > http://www.freeciv.org/lxr/source/server/gotohand.h?v=cvs#L38 > > It's intended use was to indicate to the warmap generator that a location > has already been processed. However, by virtue of Dijkstra's algorithm, > such an occurence is quite rare. This is probably the reason why it is > not quite used (although initialised) in the currecnt code. I looks like I miss something here: returned is malloc'ed, memset'ed and tested but NEVER set. So the this variable is useless. If you also come to this conclusion I don't understand why you made test runs? > I tried following things: check for "returned" disabled (that's current > code), Ok. > check enabled So what you think is the correct usage of this field? > and also plain removing all mentions of this > "returned" field. The savegames from the runs were identical in all > cases > (expected) You mention above that such "occurence is quite rare". > . The results of timing the test runs were: > If I interpret the results correctly (looking at u time), enabling the > check sometimes gives an improvement; > removal of the field gives stable although little improvement, it also > reduces volume of the code and memory used, so I suggest that it should > be applied to CVS asap. In its current form it is useless. I also vote for removing. (Mhh this is the second code construct in server/gotohand.[ch] which has no effect). Raimar -- email: rf13@xxxxxxxxxxxxxxxxx 1 + 1 = 3, for large values of 1
|