102185 - NumberFunctionCall::evaluate needs cleanup

Status: VERIFIED FIXED

(Core :: XSLT, defect)

Description

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

NumberFunctionCall::evaluate could defenatly do with some cleaning up. round() 
is broken on windows and the implementation of sum is suboptimal. Some of the 
function also needs some NaN love

Comment 1

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Attachment: this should do it

Comment 2

[Axel Hecht]

As you whack it hard, whack it harder:
Could you add a modeline? and /** **/ -> /* */? With * in the lines.
Have CEILING, FLOOR, ROUND and SUM check their input parms before the switch
and handle NUMBER separatly, and recieveError on wrong argument count.
Hrm. recieve. bah. receive. But that's a different error.

Maybe you could do the evaluateToNumber, NaN, Inf foo for CEILING, FLOOR and 
ROUND in one piece, too?

Axel

Comment 3

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Done, I also cleaned up the NumberFunctions enum (only .h file changes) and 
renamed type->mType.

I didn't do the "evaluateToNumber, NaN, Inf foo for CEILING, FLOOR and ROUND in 
one piece" since IMHO the code obfuscation by the amount of |if|s was bigger 
then the codereuse

Comment 4

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Attachment: ver 2

Comment 5

[Axel Hecht]

Comment on attachment 52560 [details] [diff] [review]
ver 2

r=me if you add the modeline 
to NumberFunctionCall.cpp.

Axel

Comment 6

[Mike Shaver (:shaver emeritus)]

Comment on attachment 52560 [details] [diff] [review]
ver 2

sr=shaver

Comment 7

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

checked in. Thanks for the reviews.

Comment 8

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

verified