Strange error on Chain to LF

[K.Forsythe]

Doesn't your field definition for APPNAME belong between the record and key definitions? I have never seen a file coded in the sequence you have it in, but perhaps it works that way? Verify that you defined ThisApp as 20 alpha, and not 2.0 decimal. Let me know if that helps. Kevin

[EDIguru]

Note the text above the file defs. It is not all one sequence. You are looking at the physical file definition. And then the logical file definition. It's old DDS. (I plan on converting to SQL tables later. The file was cloned from an existing DDS file). Yes, the field is actually alpha 20. I've accessed the file many times with upddta and entered test data into it.

[K.Forsythe]

Perhaps there is a read trigger that is failing when an attempt is made ot read a record. When the chain fails, the read trigger is not executes, and no error occurs? You should be able to find any triggers with DSPFD. Kevin

[EDIguru]

There are no triggers on the file. I created both this file and the file I cloned it from. It's just a simple control file that will feed data to the program for various processes that are application specific.

[K.Forsythe]

Well I am out of good ideas... If you want, bundle everything into a save file, download it to your pc, and email it to me at... Kevin.forsythe@dmcconsulting.com. I'll take a look at it and see if I can puzzle anything out. Kevin

[EDIguru]

I am trying a couple of things. If that doesn't work, I'll email it to you. Thanks for your time. This is really a strange one.

[Guest.Visitor]

EDIguru wrote: > I have a logical file keyed by a 20 byte alpha character. I pass a > parameter to the pgm, 20 alpha, and I am getting a strange error > (MCH3601)in debug on the chain to that file that says > pointer/parameter not set. And in the dump listing, what is the value of Appname? Bill

[EDIguru]

The value in the dump is exactly as passed on the call parameter, which matches a value that exists in the file key. I just tried something. I whittled a copy of this program down to just a small main section with only a chain to the file and a few minimal lines of code, removed everything else out of it. It has several copybooks in it to production modules that do API calls. (In fact this entire program is a clone of a program I wrote that worked in production). With only the bare essentials in the program though, it chains to the file successfully with the parameter passed. In debug of the original program it specifically errs out on that chain and positions the cursor on the F spec for the file.

[K.Forsythe]

Some optimization options mess up the debugger's ability to identify te line numbers in err.

[EDIguru]

I haven't changed anything from default settings. Is this under COMPILEOPT if I prompt on the compile? That is set to *NONE. But it does seem I am not getting the correct line number. I'm thinking it could be the copybook data in the D specs possibly throwing it off.