Re: Data Structures for IO (Re: overlaying arrays in v5r1)

As Jon noted, it's fixed for display files. And as far as I'm concerned, it could be as simple as allowing *BOTH (as opposed to *ALL) to be used for both input and output but flagging the keyword as an error if the two formats (input and output) don't match exactly. I know there's no such thing as an "easy" compiler fix, but that's seems to me to be a relatively small investment for a pretty significant ease of use payback.

Joe

On 1/22/2014 9:21 AM, Brian May wrote:

Even though I have never seen it used in practice, fields in a logical files can be defined as input only by placing an "I" in position 38 of DDS. If I am not mistaken, this is the reason that DS IO was implemented in this way. This makes it possible for the input buffer and the output buffer to have different formats. When using DSs for IO, RPG moves the DS into the buffer as a whole. In this scenario, using the same DS would cause data corruption.


On 1/21/2014 11:33 AM, Joe Pluta wrote:

Ugh. I've just had to rip data structures out of a whole bunch of
programs. It's not that I'm against the idea in and of itself, it's
just that it's not a wholesale rip and replace, for several reasons.
Here are two of the biggest:

1. You need different data structures for the input and output.

I'm also somewhat annoyed by this. If I have a physical file with a single format, can someone explain why I need to have both an *INPUT and an *OUTPUT DS? There must be something that I'm missing and I'd like to know...

Sam