Re: Adoption of RPG OA (Summary)

Yikes! I hope Larry isn't getting slammed for being confused for me! We are both so "snip"pish! (bad pun, came from the snips he and I did in the emails we responded to!)

Anyhow, you ask, shouldn't an RPG developer know how to do a CALL? Of course. OA does not remove the need for that skill. There are things that OA is not good for - like a CREATE TABLE on a remote database. I can conceive of a way to kind of shoe-horn this into an IO model, but it might be best to use procedure calls for this. In a very general sense, OA is good for things with well-defined data structures and not especially good for command-level interactions.

I like your questions - I hope that you are getting some answers that help. Personally I see Open Access as an option in situations that can be treated as native IO. The whole thing came out of an ISV requirement to be able to use native RPG opcodes to work with browsers. Now maybe it could have been new opcodes. IBM chose to make a generalized technology, that opens up more than just interactive interface modernization. For example, I can write a handler for remote database access that uses opcodes that every RPG developer knows how to use, and knows how it behaves. The only thing the developer needs to provide is an IP address for our PC middleware server, and an ODBC connection string for the database, such as Oracle or MySQL or whatever has such a driver.

I think, and some others I have talked to seem to think, that there are lots of ways to use this new technology - with the goal of concealing complexity for developers. Some seem to want to intensify the level of complexity - maybe not on purpose, but they have argued that developers should not be allowed to "languish" in "old-fashioned", "outdated" methods. (That was my emphasis and choice of words - it has been stated differently.) YOU have not said that, as far as I've seen.

Still, if you can give a developer a tool to get his or her work done faster, and more easily maintained, I think that's a good thing.

Admittedly, OA is a thin veneer over API calls. But then so are lots of things we write in service programs.

I have to admit to doing happy dances when I pull data from SQL Server running on my laptop in Minneapolis, into an RPG program running on an IBM i on an LPAR in Dallas. And it is so blasted clean-looking, so simple, so basic. I guess I really do need a real life, eh? Hey, when I want real excitement, I WRITE a record into a SQL Server table!! Then watch me dance!

Regards
Vern

On 3/19/2011 11:57 PM, Joe Pluta wrote:

On 3/19/2011 9:27 PM, Larry Ducie wrote:

Hey Joe,

That was me, not Vern.

Yeah, I get confused with who is quoting who, especially late at night
and on the weekends<grin>.

I think this situation is good for an OA handler because we may not necessarily want to (or even be allowed to) own the co-ordinate data. It may be data we need only on an ad-hoc basis. In that case, it makes sense (to me) to only get the data when it is required and let the provider (Google in this instance) manage the overhead of storing it and keeping it current. The beauty of the OA approach is that it provides a virtual table to the developer. The chain mechanism is well known to RPG developers and has expected behaviour. There is no learning curve at all for the consumer. If you can't handle a chain you shouldn't be writing RPG. The chain is designed for "random access" of data. In this case the data happens to be stored in the cloud. We could use this mechanism for many other sets of build-on-demand data. I could easily see RPG applications accessing data stored on the network via chains and read loops.

Although I endorse RPG OA in this instance I am mindful of the dangers of accessing critical data in this manner. RPG developers are not designed to worry about such things as the database not being available and may not code accordingly. Accessing data from a network resource is another matter entirely - an arena C programmers were designed for. :-)

But why OA rather than, for example, a read trigger? As I said, I'm
just gathering opinions. And just to be a little bit of a devil's
advocate, shouldn't every RPG programmer also know how to program a CALL?

<snip>
Oh, and OPM is not even in the picture - handlers are not even allowed
in OPM source - only ILE.


I think Larry meant "old programming model" in the sense of using RLA opcodes to call programs rather than using procedures or calls.
</snip>

That was Henrik, not me :-)

Henrik was talking about the "OPM (Old Program Model) and OPT (Old Proprietary Thinking) - very typical for IBM I programmers/companies."

See? I can't keep track of the players without a scorecard :)