SAMBA 2.2.8 misc issues.

[John E. Malmberg]

In article <CF0913E9C3D53D4E94DE47FDE25C633D048FD453 at hermes.cofiroute.com>,
  COLLOT Jean-Yves <jean-yves.collot at cofiroute.fr> writes:
>
>> 3. There seems to be some issues when running utilities from an account
>> that has more than tmpmbx and netmbx, but less than all privileges.  It
>> causes diagnostics about SYSLCK being missing to be generated.
>
> I thought that this problem was fixed since the 07-Aug-2003 version.
> Apparently, I was wrong.
> I can't be sure, but I don't think that the problem lies with the method of
> determining if the user is root or not. The SYSLCK privilege, as far as I
> know (and as far as I wanted it to be) is only needed by SMBD.EXE and
> NMBD.EXE
> Anyway, I need more information about that : which utilities are concerned;
> which specific command(s), if any; which additional privileges (more than
> tmpmbx and netmbx) ?

As I recall, it was the SMBCLIENT and SMBPASSWD utilities, and it was when
the SYSPRV privilege and just NETMBX, TMPMBX were active.

I needed to turn on the SYSPRIV privilege to adjust the smb.conf, and forgot
to turn it off before running the utilities.

I do not think it is something that the majority of users will encounter.

>> 4. I strongly recommend not using the /VAXC qualifier in compiling.
>
> There was already a couple of messages about that.
>
> I have not changed my mind: I don't agree with that recommendation, and with
> the arguments you give to support it. I guess we've had different
> experiences, and that yours and mine lead us toward different conclusions.
>
> However, even if I think it's not useful, it cannot add problems to remove
> that option, and I guess most of the Samba/VMS users will be happier without
> it.
> So, let's not mind about who is right (may be both of us, depending on
> circumstances) : the SAMBA UNIX code is ANSI compliant (I'll add "almost",
> if you don't mind), so it's not a such a huge work to remove that "awful"
> /STANDARD=VAXC switch, and replace it by the /WARN=(ENABLE=LEVEL4,QUESTCODE)
> you propose. Actually, the full option I am working on currently is
> /WARN=(ENABLE=(LEVEL4,QUESTCODE),NOINFORMATIONALS), and it will be used in
> the next release.

The /WARN=(ENABLE=LEVEL4, QUESTCODE) turns on a higher standard of testing
than is required by ANSI, or is done by the UNIX team, and as such it has
exposed bugs that were latent in their code.  It does generate some addtional
noise messages for issues that probably do not need to be addressed.

Again, I recommend not suppressing the informational messages as a
compiler option.  Several of the informational messages will point out
critical programming bugs, that while being legal C code will result
incorrect program operation.

SAMBA is a complext program, and as far as I know, no one has ported the
verication tool to OpenVMS, or has run it from a LINUX system against
the any OpenVMS port to verify that all the functions work correctly.

I would recommend using the "#pramga message disable xxxx" in the config.h
or equivalent file instead for the specific messages that turn out to be noise.

Also redefining the symbols for LINK and CC break building of SAMBA on IPF with
the cross compilers.  I am probably one of the few affected by that.

-John
wb8tyw at qsl.net
Personal Opinion Only