www.justiceforchandra.com

[rd]

Here's a CGIDEV2 parm handling demo I just wrote, HELLOPARMS. It's the first step to an HELLOWORLD as I get my feet wet.

http://code.google.com/p/rdwrites/downloads/list

The download is helloparms.zip

Everything in this can be found in CGIDEV2's TEMPLATE5 demo.

I didn't do anything fancy concerning looking for parms. I just look for two parms called mode and count as in:

iseries_url/helloparms?mode=visitors&count=32000

This is my first CGIDEV2 web page. The basics of putting out an RPG CGIDEV2 web page are lightning fast, development and execution.

HELLOPARMS can be run on my server with http://173.8.38.42:2700/iseries/helloparms?mode=visitors&count=32000

(mode and count can be given any values or one or both dropped off the URL as parms and resubmitted from the Go bar)

The HELLOPARMS CGIDEV2 demo source code can be viewed below. I will be releasing hopefully more interesting and useful stuff as I go along.

I want to add my vote of thanks to the CGIDEV2 team and to the IBM RPG compiler developers. I am continually surprised and pleased at the power of the RPG language and API's and now more so with the CGIDEV2 interface, and that's directly attributable to the above mentioned people.

rd

[rd]

[rd]

[rd]

a poster wrote: On some rare occasions my CGIDEV2 program crashes in the WRTSECTION module when performing CLRHTMLBUFFER with the message:
The pointer parameter passed to free or realloc is not valid.

I took a quick look at the CGIDEV2 code (XXXWRKHTML in QRPGLESRC) and can offer a couple of comments in lieu of Giovonni or someone else knowledgeable on it has time to reply.

The specific line that fails is a realloc of memory if buffer size exceeds initial alloc of 100,000 bytes. The buffer pointer is set in InitHTML and when STDOUT is done. My guess is this is triggered under some possibly rare condition where the buffer accumulates a lot of data and no output is done with it. The extreme case of this is if buffer exceeds 16 MB, the code says it will crash. But I'm thinking more of something getting out of synch when no STDOUT's are done with data exceeding ~100k. (wrtBuffer, wrtSection, etc.)

I compared to my use of clrHtmlBuffer and use of wrtSection in my small demo program HelloParms in Google Code project rdwrites. I conditioned the clear buffer with a check if anything was in the buffer which eliminates an unnecessary buffer reset (although if less than 100k looks like it's next to no code executes) and cuts down on accessing an invalid buffer (the assumption being that it is more likely to be valid with a counter saying there is something in it, as in:

// clear the HTML buffer
if getHtmlBytesBuffered > 0;
clrHtmlBuffer();
endif;

The other point is that I don't clear the buffer before each wrtSection, just once upon entry. It seems to me that just letting the (pretty complex) code do its thing and reset pointers etc is the way to go. Clear buffer is only necessary if a previous process ended abnormally leaving the buffer unflushed.

This at least gets me thinking about this stuff. I am trying to make time to take next step on it. I know a little bit more now than I did before your question.

rd