Re: CHAIN on a Partial Key

Sorry, Charlie... I use logical files with partial keys frequently. I have never seen "strange things" happen. They work extremely well and efficiently. And in 30ish years of doing this stuff, I never met another programmer that didn't understand them.

Anything you post here, there will be a handful of people who will get it wrong, and post it that way. It doesn't matter what it is -- there's always someone.

Don't let that stop you from using useful techniques!

Honestly, I think the people here who think that the sequence is unpredictable are confusing CHAIN/SETLL/READE type operations with SQL. In SQL, if you don't specify an "order by" clause, then the sequence is unpredictable. This is not true of the native I/O opcodes.

-SK




On 3/28/2013 2:51 PM, Charles Brewster wrote:

Crispin,

I am not sure if it is just self-imposed or if I was told by someone
when I was a young programmer, but I never, never, never chain with a
partial key on a logical file. Strange things can happen and as we
see from this thread, it is confusing to the programmers that follow
behind you.

Maybe it comes down to programming "style."

In my book CHAIN is reserved for when I need to update a record in a
file, the CHAIN should use a complete key that satisfies all fields
in a unique logical file.

I use SETLL and READ/READE for everything else. (Sometimes the SETLL
is enough and you never even need to do the read, performance
bonus!)

Charlie