ON EXCEPTION and STATIC CALL being mutually exclusive (was: The next question)

[Simon Sobisch]

Am 22.03.2020 um 09:53 schrieb Brian Tiffin:
> [...]
> CALL "subprog"
>
> is a dynamic lookup, and requires a library search path at run-time. 
> LD_LIBRARY_PATH and friends.  The ON EXCEPTION phrase of CALL can be
> used to test for not found.
>
> CALL "subprog" ON EXCEPTION DISPLAY "no linkage to subprog" END-CALL
>
> CALL STATIC "subprog"
>
> includes the object code in the binary, so no need for run-time
> searching, and no chance of firing an ON EXCEPTION clause.  We actually
> made that an error, and I'm protesting...  it should be a warning,
> otherwise you need to gut a lot more code to run quick tests between
> STATIC and normal CALL, and so what if the exception never fires (while
> developing).

That is a very valid point. Another one is that cobc will ignore a
static call convention flag if there's an ON EXCEPTION clause like

  CALL my-convention "some" ON EXCEPTION DISPLAY "can't find the module"

or

  >> CALL-CONVENTION STATIC
  CALL "some" ON EXCEPTION DISPLAY "can't find the module"

with a report to the programmer

  prog.cob:6: warning: STATIC CALL convention ignored because of ON
EXCEPTION


So instead of raising an error

  CALL STATIC "some" ON EXCEPTION DISPLAY "can't find the module"

as

  prog.cob:5: error: STATIC CALL and ON EXCEPTION are mutually exclusive

the 3.1 release will say

  prog.cob:5: warning: ON EXCEPTION ignored because of STATIC CALL

Checked in with r3506, will be merged to trunk later (there will be some
minor merging before the 3.1rc in any case).

Simon