Last modified: 2011-04-14 15:12:56 UTC
{{#expr: 45 round 2}} returns "45" instead of "45.00" which would be the expected behaviour and useful for aligning numbers.
Not really "broken", all expression results COULD go through the same normalization at the end, and that would strip trailing or leading zeros, force scientific format below E-5 / above 1E12, etc. Oddity, "round" does return -0 in some cases, so if you'd want normalized results (the opposite of your bug report) you have to add 0 to get rid of the bogus minus sign in -0. I didn't look at the new "formatnum" magic word, does it help?
If the number of decimal places is given, it should round to that many places to make it clear that the number is given with that precision. This is broken behavior.