Does it perhaps consider the first field blank and chain with that as the value?

Danny, is there a question in there somewhere?

All you're experiencing is the enhancement to the CHAIN opcode in free-format.

In Free-format, the KLIST/KFLD statements are no longer necessary. You use what's called "ad hoc key lists" instead. To do that you enclose the key fields that would normally appear on the KLIST/KFLD statements in parentheses. The keylist has to contain a field count that is less than or equal to the number of key fields defined in the file.

You can also specify %KDS( dataStruct : 3) and the compiler will use the first 3 (in this example) subfields of the specified data structure as the key fields. You can put any number you want in the 2nd parameter, again as long as it is less than or equal to the number of keyfields in the file.

It "looks like" you are CHAINing with the first two keys of the file and the third key (RPDOC) isn't being used by the CHAIN.

Danny, no you can't chain with key 2 and 3 while ignoring key 1.  The code you've shown is chaining with wrknum and vndnum# as the first two keys.

You're obviously in JDE and F0411 is the Accounts Payable Ledger.  Field RPRMK is a "remark" field of 30A, and RPAN8 is the "address number" and is 8,0.  RPDOC is the "document number" and is also 8,0.  So your logical F0411LAA is by remark, address number, and document number, where the program is using the vendor number as the address number in question.

But are you sure that F0411LAA is the way you show in the beginning of your post?  If so, then yes the code does seem to chain with wrknum in the remark field and then vndnum# as the address number, and not specify the document number.

Something somewhere is not quite right it seems.  Maybe either the logical F0411LAA is not laid out the way you showed, or the code may be chaining to the wrong logical.  Or wrknum and dapnum are just very poorly named and really do contain the remark?

Is this an INPUT F-Spec and not an UDPATE? If so, I *always* code chain(n) instead of chain (thanks GC). Yes, it generates a compiler warning. Why do I do this? Because as soon as the next programmer changes the IF to UF, adds an UPDATE statement somewhere in the code, compiles, installs it production, your code breaks with record locks out the wazoo. THEN after they get 50 phone calls, they check the source code back out and add the (n) everywhere. So might as well do it up front.

Chris Ringer

Ringer... also if someone adds a shared open somewhere up stream, your file might suddenly find itself an update file, with no clues in program of the problems that could transpire.