Re: flexfax: SysV configuration patch

[Robert Colquhoun <rjc@trump.net.au>]

At 19:45 18/11/98 -0500, Nico Kadel-Garcia wrote:
>> The patch you posted has an implicit dependence on a patch i wrote for
>> inclusion with the RedHat binary distribution. ie it is a patch on a patch
>> on top of 4.0pl2
>
>Gack. You're right. One more go-around....

Yes - these two patches really need merging though.

>> It is currently only suitable for RedHat systems.  At this stage unless you
>> are really confident do not include on other systems.
>
>And this is a potential problem: it's difficult to sort out the
>patches into "RedHat", "Caldera", "SunOS", "Solaris", etc.

That is why until yesterday it was 'unannounced' ;-)

The time and work to generalise it across all the unix platforms still
needed to be done.  It is very system dependent...which made me put it off.

>
>> Rather than a new parameter SYSVINIT_DIR or whatever, i was going to
>> replace SYSVINIT to represent the location of the file or "" if the system
>> used a BSD startup.  There would also be an additional prompt during
>> configure prompting the correct location for the sysv init script.  The
>
>Yup. I've added this. However, even a BSD machine should have the
>option of putting it in the config.local files (which I've done). And
>I've used a tweak to set SunOS up with the init script in $DIR_SBIN:
>the script is too useful to faxsetup and to administrators modifying
>configurations or installations to let its installation be broken by
>hunting for the full SysV init tree.

ok -how about this: Use SYSVINIT as the location of the file, if its a BSD
system (ie SYSVINIT="") then copy the script to the libexec directory as
hylafax.sysv so that it still can be used on those systems.

>
>> sysv init script would then be installed at the make install stage with all
>> the other applications, unless obviously it was a BSD machine.
>> 
>> I was still going to make the user manually install the symlinks from the
>> run-level directorys, as it is very system dependent.
>
>Hmm. I think this is a bad idea: the possibility of errors is too high.
>So I've added defaults for SysV, and RedHat Linux, that look right.

Believe me - you cannot do this automatically every system does this
slighty differently

ie for HP-UX it is /sbin/init.d and /sbin/rc*.d and the scripts range in
value from 0 to 1000
ie S905hylafax K095hylafax, HP also put rules about what numbers the
scripts are allowed to be you are *not* allowed to include one as say
S095hylafax for instance as that number range has been reserved by HP.

Also as always it is quite possible the sysadmin has further customised
this, making it all but unrecognizable to an outsider/configure script to
make sense of.

>> I have *not* done this yet.  That is why until now i had not generally
>> released this.
>
>Oh. Sorry about this: I had added your patch to my set of default
>patches for Linux/SunOS, and included for patch writing without
>realizing that it would screw up the new patch.

It's ok, i just wish you had glanced at the rpm README file which mentions
that this patch was included first.

- Robert