scottklement.com

Glad to hear you have it working. For the path option the following works. Note the use of the element count from the psds. RPG populates this when you are parsing into an array as there is no opportunity to add a countprefix field.

dcl-ds pgmStat psds;
elements int(20) pos(372);
end-ds;

Dcl-ds ...

Look into the countprefix option. That is what you need to use to get a count of the actual number. Can’t type code on my phone or I’d try to give you an example.

If you really only want that data you should also look at using the path option to direct the parse to the bit though want.

I have no clue as to a direct answer to your problem - but I can't help but wonder if there is not a better approach. Frankly I'm amazed that the process didn't just blow up at some point.

A multi-million-row spreadsheet seems insane - nobody can usefully work with that.

Can it not be grouped to ...

Been a long time since I used the "raw" YAJL APIs. I tend to use DATA-INTO with YAJLINTO for most stuff.

That said - I seem to recall that YAJL_object_find will only locate children of the passed node. So you would need to first locate the "data" node and then from there locate the child "jobs ...

I know you have been working on an alternative approach - although how SQL helped I'm not sure - but as far as I can see your second approach works just fine.

ctl-opt BNDDIR('YAJL')
OPTION(*SRCSTMT:*NODEBUGIO: *NOSHOWCPY);

/copy yajl/qrpglesrc,yajl_h

Dcl-S docNode like(yajl_val);
Dcl-S node ...

CCSID 65533? That's pretty unusual - in fact I don't think I've ever seen it before. According to IBM it is only for user-definable code sets. Which would indicate that the company is using their own conversion tables. Which might help explain this.

As to DATA-INTO, simply point Scott's YAJLGENR4 ...

It looks to me as if you are mixing up your APIs.

The second parm for http_get is the filename where you want the retrieved response placed. So hardly any wonder that you are being told the file doesn't exist because assumedly it is all blanks.

Maybe you need http_req?

If you mean has anything changed with XML-SAX that would help with this the answer is no - nothing I can think of.

Only thing I can think of is to do a scan/replace operation. But that will take time so If the failure is infrequent, better to trap the error and run the scan/replace before a retry.

Are you the person who posted this on FB? I answered it there.

And you’re being rather impatient - bouncing the question after 3 minutes is a bit fast!

Looks to me as if Should be Without that you are writing the varlen's length as part of the data and it is probably translating as blanks.

P.S. Why are you bothering to code CallP? It is not needed.