[VELOCITY-242] [PATCH] Decimals in velocity

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.5
• Component/s: Engine
• Environment: Operating System: other; Platform: Other
• Bugzilla Id: 25874
• Assignee: Unassigned
• Reporter: Peter Romianowski
• Created: 03/Jan/04 12:44
• Updated: 11/Feb/11 11:23
• Resolved: 08/Mar/07 00:07

Description

Hi,

finally we got it together - the infamous "decimal number patch" evolved
and here is the (hopefully) final patch to enable decimal numbers (and
byte, short, long and BigInteger btw.) in velocity. This patch contains:

• Decimal literals (by Will Glass-Husain)
  Now you can type things like #set ($foo = 3.14)
• All arithmetic operations (+ - / *) now work with all Number-types
  available. Modulo (%) works with all except BigDecimal (there is no
  such method
• All comparators (< > == != <= >=) work with all Number-types
• A new type called TemplateNumber (o.a.v.util.TemplateNumber) is
  available which is treated as a Number in arithmetic operations and
  comparisons. (by Will Glass-Husain)
  I am not sure about this one though. But Will had good reasons to
  implement it - I think I just need some kind of "refreshing" on that
  

The patch is against the current head (Geir, you gave me a hard time
during the last days, since you changed the Parser.jjt quite often, so
that I had to incorporate the changes quite some times

Additionally some implementation details:
-----------------------------------------

All aritmetic operations and comparisons are delegated to a new class
(o.a.v.runtime.parser.MathUtils). All operations are done using
primitives (as far as possible) and NOT BigInteger / BigDecimal like the
first proposal did. I added overflow checking all along the way (where
possible). In addition to that I removed the "paradigm of the smallest
type" where add (new Integer(1), new Integer(2)) would result in a
Byte(3). Now types are preserved as far as possible (overflow, mixing of
types). You find a lot of documentation about that in the javadoc
(MathUtils.java). Finally, overall, there should be no performance drawback.
There are also some new testcases and existing ones are updated (mostly
template based like math.vm).

So we (I hope Will doesn't mind me talking in his name) think this patch
will increase the usability of velocity a lot and hope that the others
like it too.

Open for discussion...

Regards
Peter