Re: Pushing XML out to a browser request

Lots of great suggestions coming in. Many thanks!
We are definitely going to go with a varying-length variable and a
reduction of the %trim occurrences.
Also looking into some of the other suggestions.

I will try to post back with any results we get so that everyone can
benefit from any gains we make.





From: Vernon Hamberg <vhamberg@xxxxxxxxxxxxxxx>
To: "RPG programming on the IBM i (AS/400 and iSeries)"
<rpg400-l@xxxxxxxxxxxx>,
Date: 10/03/2013 09:18 AM
Subject: Re: Pushing XML out to a browser request
Sent by: rpg400-l-bounces@xxxxxxxxxxxx



+1 on this - I remember working on a program that built HTML with lots
of lines like this -

htmlfixed = %trim(htmlfixed) + 'another long ugly string of HTML over
lots of lines in RPG code';

htmlfixed was, obviously, a fixed-length variable.

When I did this with a varying-length variable, the web page noticeable
snapped.

htmlvarying = htmlvarying + 'the same ugly long string - anyone up for
CGIDEV2?';

And htmlfixed was probaby 32K - there are probably ways to optimize
%trim, but conceptually, don't you have to scan from the end back to the
first non-blank? For every time you append some data? Varying-length
variables tell you where that end is - ba-da-bing easy!

Vern

On 10/2/2013 5:13 PM, John Yeung wrote:

On Wed, Oct 2, 2013 at 4:47 PM, Briggs, Trevor (TBriggs2)
<TBriggs2@xxxxxxxxxxx> wrote:

Are you doing the:

@Xml = %Trim(@Xml) + ...

multiple times on each build?

I'll bet this is a major (if not the major) factor slowing him down.
About a year ago, there was a thread on this list titled "XML
performance issue" (starts with
http://archive.midrange.com/rpg400-l/201209/msg00135.html). It came
up during that thread that lots of trimming and joining of large,
nonvarying strings is a big performance hit. If @Xml is a
fixed-length string destined to become a CLOB, it sounds like that's
one likely place to get speed improvements.

John