Re: %trim expensive-ness - was (RE: right$)

"Bartell, Aaron L. (TC)" wrote:
>
> Hans,
> >Oh yeah, that first EVAL with the %TRIM is pretty expensive.
>
> In my CGI programs I use %trim() extensively, but I don't necessarily need
> to do a %trim() but instead I could do a %trimr() because I just need to get
> rid of the right most characters 90% of the time. How much do I gain, if
> anything, by doing a %trimr() vs. a %trim()?
>

If the data is already left-justified, then by using a %trim, your
program is a bit bigger than it needs to be and on the %trim you have to
go through a bit of extra code: the loop setup and the test of the first
byte.  Not enough to worry about if there's a possibility of unwanted
blanks at the beginning, but otherwise, using %trimr is both more
precise and causes more concise code generation.

%trim-heavy code would probably be improved in many ways
(readability/maintainability, performance, program size) by using
varying fields.  Even if your input parameters are not varying, if you
copy them trimmed to varying fields, the rest of your code will be
simpler and faster.  (It won't take many trims to make up the time for
the initial copy of the fixed field to the varying field.)

Barbara