RE: Suggestion: File to use when compile

> Actually I am not necessarily in favor of the pre/post compile
> commands, the potential for people to put in commands that mess up
> something important, i.e. CLRPFM or DLTLIB, is much too great.

Someone else had concerns along the same lines.  I think something is being
missed here.  Why would commands like these even be in a pre-compiler
script?  Don't you trust your programmers?  If they can make that kind of
mistake here, then you better take the command line away from them.  And
don't let them write any CL programs.

db

> -----Original Message-----
> From: rpg400-l-bounces@xxxxxxxxxxxx / Joe Lee
> Sent: Wednesday, October 13, 2004 8:31 PM
>
> I have to agree with Tony here. I think that any compiler options should
> remain in the H specs. While I agree that the pre/post command option
> would take care of this and would ensure that any new compiler options
> were supported without depending on IBM, I would much rather have the
> consistency of having all of the compiler options in the H specs or none
> of them, and I don't see IBM breaking the current behavior of the
> compiler by removing that support. I guess I have to ask the question,
> is team responsible for the compiler commands different from the team
> responsible for the compiler/precompiler? If not, then why can't support
> for any new compiler options be added when those options are added?
>
> Actually I am not necessarily in favor of the pre/post compile
> commands, the potential for people to put in commands that mess up
> something important, i.e. CLRPFM or DLTLIB, is much too great. The nice
> thing about the keyword solution for the file problem is that there
> would be no need to create a temporary file before or during the
> compile. When I compile a program the only thing that should happen is
> that my program is compiled, the compiler should not be capable of
> permanently changing any object other than the object being compiled. So
> I think that there should be definite limits on what commands could be
> executed and possibly over the objects they could be executed on. That
> being said, it would be nice to be able to include commands to add
> constraints to a physical file after it was compiled.
>
> Joe Lee