Re: XML performance issue

Jon,

2 questions I'm not sure of that you *may* be able to answer:

1. Is there any difference between what is generated at a low-level (W-code
or whatever) for "A = A + B" and "A += B"? I assume not, and that the
difference is merely in the RPG, but I don't know whether the compiler can
make any optimization based on the second example.

2. If one uses %Trim on a left-adjusted field (so there are no left spaces
to trim), what's the actual run-time overhead? I'm guessing there are a
couple of lines of code which get executed, but it can't honestly be much,
can it? Sure, there's still the right-trimming to be done, but is there a
noticeable difference between using %Trim and %TrimR on a left-adjusted
field, even through thousands of iterations (especially if there is
considerable trimming to be done on the right, which is what would take the
majority of the time). I could probably test this myself, but no doubt
wiser heads than mine (yours, for example), have already done this.

Rory

On Sun, Sep 16, 2012 at 9:15 AM, Jon Paris <jon.paris@xxxxxxxxxxxxxx> wrote:

On 2012-09-15, at 1:00 PM, rpg400-l-request@xxxxxxxxxxxx wrote:

I would love to see some RPG examples of this.

Simplest one of all:

D myBigField s 20000a Inz
D myBigVaryField s 20000a Inz Varying

D data s 40a Inz('Only 12 used')

D comma c ','

D i s 5i 0 Inz

/Free

For i = 1 to 1000;
myBigField = %TrimR(myBigField) + %TrimR(data) + comma;
EndFor;

For i = 1 to 1000;
myBigVaryField += %TrimR(data) + comma;
EndFor;

*InLR = *On;

Try this and you'll see that the second (varying) version is orders of
magnitude faster. I've seen differences in excess of 1,000 times. I've also
seen versions worse than the first one in that they used %Trim in both
cases and not %TRimR.

And before anyone points it out, yes I know I probably don't want a comma
at the end of the data ;-)


Jon Paris

www.partner400.com
www.SystemiDeveloper.com