Re: [reSIProcate] representing qvalues as integers instead of floats

[Dan Weber <dan@xxxxxxxxxxxxxx>]

Rohan Mahy wrote:

> Hi,
>
> I just checked in some code in ParserBuffer.cxx that parses a q-value 
> and returns an integer between 0 and 1000.  Some examples:
>
> q-val        integer
> 1        1000
> 1.0        1000
> 1.00        1000
> 1.000    1000
> 0.0        0
> 0.00        0
> 0.000    0
> 0.8        800
> 0.80        800
> 0.05        50
> 0.050    50
> 0.001    1
>
> I think it would be a good idea to migrate the q parameter to this 
> type instead of the float type.  Some folks have pointed out that 
> using a float for a q-value results in the gratuitous addition of 
> floating point code to platforms that may not otherwise "need" it.
>
> thoughts?
It makes sense, since the standard doesn't really specify the value as a 
floating point.  Is there anyway to keep it from breaking existing 
code?  As well, it might make sense to use an enum to set the 
appropriate value.

Dan