On Sun, May 26, 2002 at 01:18:04AM +0200, Tibor Pittich wrote:
> no, i don't have two choices. if i want use only binaries and don't want
> rebuild telnet or ftp package, i _must_ have installed kerberos libs
> too. i know, this is offtopic on this thread, but how many percent users
> use kerberos authentication ? if you can reply to this, please change
> topic.

*sigh* first off you're not understanding me.  Mandrake has two choices.
Ship a kerberose telnet client or don't ship a kerberose enabled telnet
client.  If they do it has to be compiled with the kerberose libraries.

Further, the topic is still applicable.  I'm still talking about
spamassassin.  Just because you're using a poor example to argue that
the requirement should stay doesn't mean that we're not still talking
about spamassassin.

> is this anywhere strictly defined as it is interpreting by you?
> i read mdk-rpm howto and maximum-rpm doc, but there is no allusion that
> this field is dedicated to telling which package is _mandatory_.

Huh? The Mandrake RPM Howto says of Requires:
"This line was added because one of the programs included in the package
is a python script. It then needs python to be executed. You can put
optional minimum (or equal) version, for example: Requires: python >=
1.5.1"

While this is an example.  The word *NEEDS* is in the sentence.

And the Maximum RPM Document says of the Requires tag:
"The requires tag is used to alert RPM to the fact that the package
needs to have certain capabilities available in order to operate
properly. "

*NEEDS* *OPERATE PROPERLY*

Not *SHOULD* or *MAKES IT WORK BETTER*

> <b>i don't wrote</b> that spamassassin won't work without razor-agent.
> i wrote only that spamassassin works much _better_. i know, my english
> is very bad, but i believe that this my idea was comprehended clearly.

In your discussion explaning spamassassin scoring certainly does imply
that you won't catch very much spam with it if you don't use it with
Razor.  Thus in my opinion you were imply it wouldn't work *right*.  At
any rate I am certainly aware of the requirements and functionality of
spamassassin.  I think my original email in the thread sums that up just
fine.

> it is configurable.. or you can use --nodeps. look at my example about
> krb5-libs. it is not configurable and i MUST have kerberos libs.

You can only choose to have kerberos in telnet at build time.  Not run
time.  It links with the krb5-libs.  Thus if you enable it at build
time, even if you turn off kerberose authentication at run time, you
still have to have the libraries.  That's a function of the way compiled
languages work.  spamassassin is mostly written in Perl.  Perl is not a
compiled language and is more flexible about library requirements.

As Murry as already pointed out it's stupid to have to install something
with --nodeps when the requirement you're trying to get around isn't
really needed.  And there are *VALID* requirements upon spamassassin
that you wouldn't be warned of if you use --nodeps.  E.G. perl itself.

> but this is not problem in razor agent ;) this problem related to
> default config file. imho, better is disable this functionality by
> default. on this point you have right.

I don't see any reason to disable Razor functionality in SpamAssassin by
default.  That's inconvient to the vast majority of the SpamAssassin
users.  But I also don't see a reason to require Razor to be installed.

> fundamentally question is: where is authoritative description about
> "Requires" and what tells about it. if it exists and you have right in
> your interpretation, and my lobbing about "Requires" for spamassassin and
> razor-agent , please sorry for my mail and don't throb my head :) i was
> thinking, that this forum is right place for discusion about this.

You pointed out two pieces of documentation which clearly explain what
requires means and are supportive of my opinion.

Really it seems to me that you don't understand the nature of library
dependencies in programs written in Perl or C.  Which would imply that
you probably shouldn't be commenting on such matters.

-- 
Ben Reser <[EMAIL PROTECTED]>
http://ben.reser.org

We tend to see all wars through the lens of the current conflict, and we
mine history for lessons convenient to the present purpose.
- Brian Hayes

Reply via email to