Patch for making 2.2.1 compile Under NCR MP-RAS (SVR4)

[John E. Malmberg]

On Thu, 12 Jul 2001, Seth Mos wrote:

> At 12:00 12-7-2001 -0400, John E. Malmberg wrote:
> >On Thu, 12 Jul 2001, Seth Mos wrote:
> >
> > > Hi,
> > >
> > > I have attached this patch for compiling samba-2.2.1 under NCR MP-RAS.
> > > The patch is not intrusive and only changes the words TRUE and FALSE to
> > > True and False because the C compiler on this system gives the "Identifier
> > > is undeclared" message.
> >
> >4.9 K Application / Octect stream attachment?
> 
> Oops the mailer messed up.
> I could send you the patch in the body.

Actually there is a different mailing list for patches to be sent to.

Patches should be in diff -u format and sent to samba-patches(@)samba.org.

They also should be done against code that is obtained from the CVS
HEAD branch from the CVS server.  See http://www.samba.org and follow the
links.
 
> >I am showing that TRUE and FALSE are not defined on Compaq C if you
> >specify strict ANSI compliance.
> 
> Ok, so I how do I switch it off then?

It is not a matter of switching compliance off, those extensions must
also be know to your NCR compiler.

> Note: 99% percent of the code uses True and False.
> It may just have been some pieces of code that slipped through.
> It are only about 5-6 occurences so they seem to be typo's.
> gcc does not complain and just compiles the thing.

I am not on the SAMBA team, so I do not know if there is a specific
effort to not rely on common non-standard extensions or not.

If there is, then I have some more patches that I can send in where
they are requiring the presence of routines that I can not find
documented in The Open Group's online reference, or they are using
routines that either The Open Group or the LINUX documentation has
declared obsolete, and recommends using a different routine.

But these are all minor issues, and not showstoppers.

> >The proper fix would be to have these symbols defined in CONFIG.H if the
> >configure script does not detect them present.
> 
> It does define False and True.
> 
> >This of course means that the configure script should also be be patched.
> 
> I don't see how this would matter.

It all depends on what degree of portability or Coding standard that
the SAMBA Team is aiming for.  I have not seen any specific document, but
then again, I have not looked.

The problem with something small like this, is that so many programmers
are used to those symbols being there, that the simplest route to long
term maintenance is to insure that they are there.

Otherwise, you might fix it this time, and someone could add a new module
to SAMBA with the same bug.

The only way to stop that is if the SAMBA Team had some sort of automatic
tool that insures that the code meets their standard.

Sometimes it is just better to go with the flow, instead of trying to get
things to be perfect.

-John
wb8tyw at qsl.network
Personal Opinion Only
 
If you like the challenge of building /porting SAMBA, you can join me
on OpenVMS.  See http://www.montagar.com/hobbyist for information how you
can get free (personal use only) licenses and stuff to get you started.
Follow this thought up on the samba-vms(@)samba.org list, and not the
samba-technical list.