Le lundi 12 septembre 2005 à 22:38 -0700, Nathaniel Smith a écrit : > On Mon, Sep 12, 2005 at 08:38:33PM +0200, Beno?t Dejean wrote: > > here's a possible fix for this bug : i've added ticker::id to store the > > unique identifier of each ticker. using N_() then _() would be another > > solution, but i think it would be harmful to do so. That's why i stored > > id and name seperately. ticker::id gets initialized with a > > locale-independant string. As i've added ticker::id, i've turned the > > tickers set from map<string, ticker*> to set<ticker*> which make more > > sense. Notice that ticker::id and tickers as a set are only used for > > sanity checks. > > Why provide a string ticker::id at all?
for sanity checks only. > It's just something to > copy-paste error on... why not identify tickers by their memory > address, or some unique id generated at construction time? as far as i understand the author's intention, each ticker has a unique role. You can't get this semantic with memory address or generated id. Do we really need all these checks ? i'd rather replace the set<> with a simple vector<> and remove all tests :) _______________________________________________ Monotone-devel mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/monotone-devel
