Bug 113875

Summary: /etc/nsswitch.conf contains refrence to non-installed lwres module.
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Miguel de Icaza <miguel>
Component: BasesystemAssignee: Thorsten Kukuk <kukuk>
Status: VERIFIED INVALID QA Contact: E-mail List <qa-bugs>
Severity: Normal    
Priority: P5 - None    
Version: Beta 3   
Target Milestone: ---   
Hardware: Other   
OS: All   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---

Description Miguel de Icaza 2005-08-29 14:38:06 UTC
The /etc/nsswitch.conf file contains a description of the various values
that are allowed in the file.  There is a new value called "lwres" which is not
documented in the file nor the manual page.

The line looks like:

hosts:          files lwres dns
Comment 1 Thorsten Kukuk 2005-08-29 15:11:34 UTC
It is documented in the nss_lwres package.
Comment 2 Miguel de Icaza 2005-08-29 15:29:40 UTC
The issue is not whether there is some obscure package that documents it, but
the fact that:

* /etc/nsswitch.conf has a descriptive information of stuff of everything except
lwres.

* The man page for nsswitch.conf (which is what I would be man()ing for) does
not include this information
Comment 3 Miguel de Icaza 2005-08-29 15:30:33 UTC
In addition, the nss_lwres package is not installed by default on my machine, so
having something in nsswitch.conf referencing it seems like another bug as well.
Comment 4 Thorsten Kukuk 2005-08-29 15:44:43 UTC
(In reply to comment #2)
> The issue is not whether there is some obscure package that documents it, but
> the fact that:
> 
> * /etc/nsswitch.conf has a descriptive information of stuff of everything except
> lwres.

This is wrong. nsswitch.conf has a desriptive information of stuff coming with
the same package as nsswitch.conf, and nothing else. Everybody can write its own
nss module and put it to his product, you cannot document this.

> * The man page for nsswitch.conf (which is what I would be man()ing for) does
> not include this information

Correct, because it can also only document about what it knows, not about what
somebody else can program.

(In reply to comment #3)
> In addition, the nss_lwres package is not installed by default on my machine, so
> having something in nsswitch.conf referencing it seems like another bug as well.

This is no bug but a feature.
Comment 5 Miguel de Icaza 2005-08-29 18:33:29 UTC
nsswitch.conf as installed in Beta3 *does* reference lwres, it is not something
I just made out.  So if we are going to distribute something that *references*
by default lwres in the default configuration we:

* Should add a dependency on said module.
* We should have a terse explanation.
* We should have it documented.

Maybe you want to close the bug because it is not important to you, feel free to
make the priority "Wishlist", but this is a bug, not an invalid problem

The purpose of the documentation is not to be rigid but to actually assist
someone trying to understand, fix, improve or customize an error.

For example, there is a "SEE ALSO" section that you could fill in with
information about lwres, hey, am going to give you some help:

.SH SEE ALSO
.PP
See nss_lwres(1) for information on the lwres module

Comment 6 Thorsten Kukuk 2005-08-29 18:44:43 UTC
(In reply to comment #5)
> nsswitch.conf as installed in Beta3 *does* reference lwres, it is not something
> I just made out. 

And in another configuration it referneces ldap. or db, or ....
Read again comment #4.

> So if we are going to distribute something that *references*
> by default lwres in the default configuration we:
> 
> * Should add a dependency on said module.

Read again comment #4. We will __NEVER__ do that, nss_lwres is not
even part of every product we have.
And we don't wish to have it installed as default.

> * We should have a terse explanation.
> * We should have it documented.
>
> Maybe you want to close the bug because it is not important to you, feel free to
> make the priority "Wishlist", but this is a bug, not an invalid problem

No, it is no bug, it is a feature.
 
> The purpose of the documentation is not to be rigid but to actually assist
> someone trying to understand, fix, improve or customize an error.

And the purpose of not documentation this is to have no need to support it for
7 or 15 years on enterprise products nobody wish to do.

This entry is there only to make the life of some poeple easier, not to waste
developer power on it.
Comment 7 Miguel de Icaza 2005-08-29 21:51:44 UTC
I read comment #4.

What you did not read is the fact that nsswitch.conf as shipped on Beta3 *does*
reference the non-existing nss_lwres module.

That is a bug.

The other issues (documentation, being useful, having a pleasant system) is a
different matter.  I think we should ship the best product we can, you think we
have to ship the most orthogonal system, and thats debatable, and not something
am interested in debating for long.

You dont care about the docs?  Fine.   You dont care about a polished product? 
Fine.

But the reference to lwres by default in our nsswitch.conf remains a bug.
Comment 8 Thorsten Kukuk 2005-08-30 03:59:43 UTC
(In reply to comment #7)
> I read comment #4.
> 
> What you did not read is the fact that nsswitch.conf as shipped on Beta3 *does*
> reference the non-existing nss_lwres module.

I read that. And as I already answered: This is intentional.

> That is a bug.

No. it is not. You really should let people do the job who do this some years
longer and know what they are doing.
Comment 9 Miguel de Icaza 2005-08-30 04:20:16 UTC
Failing to provide a compelling reason why this bug must be closed, your only
solution was to pull rank, or as the cool kids call it, "appeal to authority" to
make your point. 

At this point I do not expect you to explain why the default installation in
Beta3 references a module that has not been installed.  And if it has to  depend
on it why this dependency is only documented in your brain (leaving aside your
condescending remarks that have not shed a lumen of light into the problem). 

The time it took you to write close the bug all these times and to poorly argue
is probably more than the time it would have taken to close the bug.

We clearly have different standards for overall software quality.