Mailing List Archive

FSSTD is linux specific?
On 23 Feb 1997, Russell Nelson wrote:

> Ira Abramov writes:
>
> > - clean up the last few backward compatabilities with sendmail
>
> Which are?

umm, commandline options and stuff... someone mentioned a few days ago
that the commandline parameter user1,user2 on sendmail worked right (sent
to user1@localhost and user2@localhost) but in qmail's sendmail it did
<user1,user2@localhost>
stuff like that...


> > - find a nice solution for people used to using .forward
>
> .forward is badly documented. In most cases .forward can be renamed
> to .qmail.

yes, but go and send the word about it to a 1500 user system. 100 messages
will pong "how do I rename it?" and 400 more will be "what is .forward?"

I think the users shouldn't care what the mail server is. I have users
coming in and creating .forward files without asking... I don't envy
admins of big systems having to explain to every new yet experianced user
that comes through the door that .forward is dead, and .qmail is the one
to use... people like to import all their dot files and expect to go on
working as usual.

.forward doesn't have an MTA-specific sound to it (it's not .sendmail, and
I won't get more specific). why not just keep it as the default name?

better yet: have the compilation-time scripts ASK you, defaulting to
.qmail, but making .forward a possible name too. (or sticking it in the
runtime config files, add that to the "hardcoded UIDs revisited" thread)

>
> >- RPM and deb packages becoming a standard official distribution format,
> > right next to the .tar.gz file, to help Qmail spred faster
> > - stick to FSSTD and kick the binaries/manpages out of /var in the default
> > installation
>
> Kind of Linux-specific, aren't these?

FSSTD makes a lot of sense in ANY unix, IMHO. I wish something like it was
used on a more global scale. when a sysadmin skips from win95 to unix, he
remembers to flip the switch in the head, but between linux, AIX, Sloaris
and HPUX the shelles and most everything looks similar, yet the files have
to be found each time in different places...

at any rate, like many here said, /var is NOT a place for binaries for
many more than one reason. I'll symlink my installation out of there, but
it IS something to look at. I have no prolonged experiance with Unices,
but I have yet to find standard distributions, especially ones that are
freewares or GPLed and the like and are classified under "contrib" that
have the "nerve" to use /var for binaries :)
Re: FSSTD is linux specific? [ In reply to ]
>>>>> "Ira" == Ira Abramov <ira@scso.com> writes:

Ira> overflow. so now I should stick a symlink by default in all the
Ira> directories, as part of my /etc/skel ?

no. you should do what i suggested.

Ira> .qmail is a superset of .forward, is it not?

no, it's not.

Ira> how about qmail looking for ~/.qmail, then .forward if
Ira> non-existant, and then default to deliver as usual? looks like the
Ira> right "UNIXy" solution.

i told you how to do this. pressing a point, particularly with a
suggestion that has already been given as a solution seems odd.

Ira> stop thinking of Qmail as a per-host added feature and start
Ira> looking at it as a standard part of distributions if you want to
Ira> see it hit the big time... the old sendmil is in /usr/sbin on my
Ira> Linux, why not have all the qmail binaries there too?

do it if you want. my sbin directories get overwritten every time i
upgrade the os. if i want to save something it needs to be somewhere else.
btw, it's not the big time unless the vendor buys into it and then it will
be dribbled around the file system in places like /usr/lib, /etc/mail and
/var/log (or worse put in /opt).

Ira> like saying that logs go into /etc, because that is machine
Ira> specific :(

my logs are in /var/log

when someone asks djb about integrating a binary distribution and he says
nope tell us about it. until then -- well if you're a system admistrator
act like it*, otherwise hire one. check www.qmail.org, maybe you can find a
consultant.

*i've edited binaries to change poorly thought out file names.

--
paul ``what the hell is linux and why should i care'' graham
pjg@acsu.Buffalo.EDU |public keys at:
| http://urth.acsu.Buffalo.EDU/~pjg/key.html
if the above contains opinions they are mine unless marked otherwise.
Re: FSSTD is linux specific? [ In reply to ]
Ira Abramov writes:
> On 23 Feb 1997, Russell Nelson wrote:
> > > > .forward is badly documented. In most cases .forward can be renamed
> > > > to .qmail.
> > >
> > > yes, but go and send the word about it to a 1500 user system. 100 messages
> > > will pong "how do I rename it?" and 400 more will be "what is .forward?"
> >
> > And how do *you* explain how .forward works when a user asks you??
>
> I don't tell newbies about features unless they need them, that saves me a
> lot of help desk work, but if a non-newbie steps in and asks me why
> doesn't my system acknowladge his .forward file, or worse, I could have a
> user leave a .forward without verifying it works, and fly to Madagascar
> for 2 monthes, and his mailbox will overflow. so now I should stick a
> symlink by default in all the directories, as part of my /etc/skel ?

No. You should check for the existance of any .forward files in a
script run nightly, and beat up users who create them.

> .qmail is a superset of .forward, is it not? so why a new filename? my
> golden rule says: drive the sysadmin nuts if needed to get you somewhere
> better, but don't try and educate the users by force.

No, the .forward syntax is quite crufty, and not anything that should
be standardized. .qmail files are much cleaner.

> how about qmail looking for ~/.qmail, then .forward if non-existant, and
> then default to deliver as usual? looks like the right "UNIXy" solution.

Since the format of .forward files is not carefully specified, just
exactly what are you going to support? About the only way I can think
of doing it is to examine the source of sendmail and see how it
handles .forward. Dan already told you how you might write a program
which supports .forward files.

--
-russ <nelson@crynwr.com> http://www.crynwr.com/~nelson
Crynwr Software sells network driver support | PGP ok
521 Pleasant Valley Rd. | +1 315 268 1925 voice | Peace, Justice, Freedom:
Potsdam, NY 13676-3213 | +1 315 268 9201 FAX | pick two (only mostly true)
Re: FSSTD is linux specific? [ In reply to ]
c'mon russ, you're forgetting the tao of qmail. for the handful of users
with forward files just let sendmail handle it. certainly you'd like to
discourage their spread but unless everyone at the site is filtering with
filter forward files should be rare. if everyone has a forward file there
are more serious problems that should be dealt with first.

>>>>> "Russell" == Russell Nelson <nelson@crynwr.com> writes:

Russell> No. You should check for the existance of any .forward files
Russell> in a script run nightly, and beat up users who create them.

--
paul
pjg@acsu.Buffalo.EDU |public keys at:
| http://urth.acsu.Buffalo.EDU/~pjg/key.html
if the above contains opinions they are mine unless marked otherwise.
Re: FSSTD is linux specific? [ In reply to ]
In message <199702261659.LAA13240@math.psu.edu>, Raul Miller writes:
>Dave Barr:
>> There's nothing inherent with qmail that makes binaries and man
>> pages host-specific.
>
>Why do you say this?
>
>pre-compilation configuration of qmail affects both binaries and man
>pages (which includes host-specific table lookups).

Exactly, but there's nothing which inherently says that they _are_
host-specific. If you wish to make it host-specific, fine. However
my point is that they are as inherently installation-specific options
as they are host-specific. Either way, binaries don't belong in /var.
Host-specific binaries if anything belong in /sbin. However I don't
beleive that qmail binaries should be encouraged to be host-specific.
That's just bad sysadmin practice.

--Dave