RE: DDS Support

Don wrote:
>...
>I'll go one futher:  Why can't users write their own opcodes?  We saw the
>concept back in RPG 2.5 where macros in effect were copied in at precomile
>time...but, why can't I create a op code, put it in some set lib area and
>have it as a valid one that the compiler will just understand and roll
>with?

Why can't users (I believe you mean programmers) write their own
opcodes?  Well, strictly speaking, they can.  Study up on
procedures in ILE RPG.


>From what I've spoken with some of the folks on the MI400 list off
>line, creating a compiler for this box isn't that much of a chore and
>heck, why not just make the compiler available to those that want to
>customize their own compiler...yeah, I know it tends to throw standards in
>the crapper...maybe..:)

Do you mean that using MI as an intermediate language makes it
easy to write a compiler for the AS/400?  Then why do we use
W-Code today in the ILE compilers?  Unlike MI, W-Code is a
binary data stream, with an instruction set that resembles a
low-level stack-based machine.  Although MI may be a good
choice for easy compilers, W-Code is much better for good
optimizing compilers.  After working with both styles of
compilers, I can say that I much prefer working with W-Code.
Although MI is itself a powerful language that matches RPG
very well, the places where MI semantics differ from RPG
semantics make the compilation process difficult and the
resulting generated code inefficient.


>AND, then there's my MAJOR pet PEEVER with RPG...when the heck are we
>going to get multidimensional arrays!?  Prefereably ones that org:1 not
>org:0 like C, Java, etc....or better yet to appeas the unix/C weenies,
>allow a compile time option to allow the programmer to decide if they
>org:0 or org:1 (would be a NICE feature for Java and C as well!!!! HINT :)

When we started working on the V5R1 ILE RPG compiler, we sent
around a poll asking for input on what should go into the next
release.  Multi-dim arrays came up just a few items below the
items that were chosen for inclusion.  Whenever we get around
to starting on V5R1+1, we'll very likely do another enhancement
poll, and my prediction is that multi-dim arrays will finally
get voted in.

Regarding the origin, currently, RPG uses a 1 origin for arrays,
and that will not change for multi-dim arrays.  A choice of
origin is more trouble than it's worth.


Cheers!  Hans

Hans Boldt, ILE RPG Development, IBM Toronto Lab, boldt@ca.ibm.com


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---