Re: Pull 638 - floats in user-defined comparisons

On Sun, Apr 6, 2014 at 8:28 PM, Stas Malyshev <smalyshev@sugarcrm.com>wrote: > Hi! > > I looked at pull 638 and it looks OK code-wise. Are there any objections > still to merging it? It changes behavior slightly and the warning in the > docs no longer applies, but I think new behavior is actually better than > the old one as for most doubles between -1 and 1 we do not get random > sorting behavior anymore. > I'm not entirely sure if this is such a good idea, but I also can't say it's a bad idea. It'll likely break someone's code out there that relies on the existing behavior (for some strange reason). I thought it was actually a better idea to document the behavior. Which is why I added the caution in the manual. It's simpler and removes the ambiguity about what's expected from the callback. This PR just adds some complexity for little more than added magic. I'm not a fan of magic in general, but I can see how most users think of magic as being useful... That is of course, until it comes back to bite them. Not opposed to merging this, but just my 2 cents.