Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Tue, 31 Oct 2006 14:53:46 -0600

Ernest L. Williams Jr. wrote:

John Sinclair has been working with us to run alh under linux_x86_64 and
has found that the alh core dumps at run time due to the definition of
NULL that comes from EPICS BASE.

======================================================================================
dbDefs.h:#if     defined(NULL)
dbDefs.h:#undef NULL
dbDefs.h:#define NULL    0
dbDefs.h:#define NONE            (-1)    /* for times when NULL won't do
*/
dbDefs.h:#define PVNAME_STRINGSZ                61              /*
includes NULL terminator for PVNAME_SZ */
========================================================================================

What is the recommended way to move forward on this?
Could we just get rid of the definition in dbDefs.h or will that break
some legacy apps?

Notwithstanding all the language lawyer comments about how NULL should

be defined, removing the definition from dbDefs.h does not break the

build of Base (on Linux, solaris and vxWorks 5.4 at least, I haven't

checked the other archs), so I think we should remove it. According to

my K&R Second Edition, NULL is defined in stdio.h, although that might

have moved to stddef.h in more recent versions. Unless someone screams

it will be gone by R3.14.9. This doesn't mean that we shouldn't fix

alh, but having our own definition of NULL is obviously a Bad Thing anyway.

I'm about to add a response to the more detailed argument going on

between John & Eric...

- Andrew
--
There is considerable overlap between the intelligence of the smartest
bears and the dumbest tourists -- Yosemite National Park Ranger

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System