I currently have a PMR open with the support minions of our lord Ib'm for v7.1 concerning un-initialized pointer variables containing errant data at start-up time. (The problem doesn't occur on v5.4.) This has lead them to open an APAR SE45175 which they said would be closed as PE (Programming Error). This is only related to the subject at hand because at one point I let the support minion know that I wouldn't have even known about this problem if CL had some kind of *NULL special value like RPG currently has. He passed this comment to the saints of development and they responded that it actually did and it's had it since v6.1. They even included a little program I could compile to prove it.
After proving it I immediately re-scoured the "What's New in CL" sections of the Information Center for v6.1 and v7.1, but couldn't find anything. Then I searched the CL sections for both releases, but came up with nada. The support minion also searched through the Info Center and like me came up empty handed. He's since notified the powers that be to get that fixed but we might not see it documented until the next release.
I also asked around and fellow high priest Father Jerome Hughes was kind enough to pass along this link to a Power-Point presentation given by Ib'm saint Guy Vig at an Omni User Group ecumenical dinner. Discussion of the *NULL value can be found on slides 93 & 94.
In any case here's the small program that the saints of development passed onto me:
PGM DCL VAR(&PTR1) TYPE(*PTR) ADDRESS(*NULL) DCL VAR(&CHAR1) TYPE(*CHAR) LEN(1) CHGVAR VAR(&PTR1) VALUE(%ADDR(&CHAR1)) CHGVAR VAR(&PTR1) VALUE(*NULL) IF (&PTR1 *EQ *NULL) THEN(DO) SNDPGMMSG MSG('The pointer is null') ENDDO ENDPGMAs you can see *NULL can be used to initialize declarations (DCL) assignments (CHGVAR) and in conditional statements (IF).
UPDATE (10/06/10): Here's something pretty interesting. On a lark I compiled the program to TGTRLS(V5R4M0). I had to remove the ADDRESS(*NULL) parameter from the DCL statement, but other than that it compiled. I installed the program object to a v5.4 system and it ran exactly like it did on the v7.1 system. However when I FTPed the source to that system and compiled it the compile failed saying the "*NULL" was incompatible with the variable.
No comments:
Post a Comment