Re: Egads! An area in which the precompiler people surpassed the compiler people!

[Hans Boldt <boldt@xxxxxxxxxx>]

rob@xxxxxxxxx wrote:
> The SQL  SET OPTION is SQL's version of the H spec.
>
> For example:
>      C/EXEC SQL
>      C+ Set Option
>      C+     Naming    = *Sys,
>      C+     Commit    = *None,
>      C+     UsrPrf    = *User,
>      C+     DynUsrPrf = *User,
>      C+     CloSqlCsr = *EndMod
>      C/END-EXEC
>
> Here's something you can do in a SET OPTION that you cannot do on an H 
> spec.  Why?
>      C/EXEC SQL
>      C+ Set Option
>      C+     TgtRls = V5R1M0
>      C/END-EXEC
>
> Rob Berendt

Will wonders never cease, eh?

I don't know how the SQL prep works, but generally, the RPG command 
options that you can't code on the H-Spec are disallowed there 
because we're already knee deep into the compile already. In 
particular, when you specify a different release on the CRTBNDRPG or 
CRTRPGMOD command, a completely different program is invoked to 
compile the source code. Since the H-Spec's are compiled by that 
source code, it's meaningless to specify which version to use on the 
H-Spec.

I suppose when the H-Spec parsing sees that a different compiler 
version is requested, we could abandon the compile and tell the 
command to invoke a different version of compiler. But then, how 
would the compiler handle the following H-Specs:

H/if defined(*V5R2M0)
H   TGTRLS(*V5R1M0)
H/elseif defined(*V5R1M0)
H   TGTRLS(*V5R2M0)
H/endif
H*         ;-)

Cheers! Hans