Pkg exim4 users - May 2006 - question about update-exim4.conf manpage

The man page for update-exim4.conf says, in part

  If update-exim4.conf is called without the option -o, it
  checks the validity of the freshly generated configuration file and
  exits with  an error.   In   that   case,   the   new   file   is
  not  installed  to /var/lib/exim4/config.autogenerated.

Is that the intended meaning?  My guess is the intended meaning is

If update-exim4.conf detects that the freshly generated configuration
file is invalid it exits with an error.  Unless the -o option was
given, the old output file is retained.

On 2006-05-20 Ross Boylan <ross@biostat.ucsf.edu>
wrote:
> The man page for update-exim4.conf says, in part
>   If update-exim4.conf is called without the option -o, it
>   checks the validity of the freshly generated configuration file and
>   exits with  an error.   In   that   case,   the   new   file   is
>   not  installed  to /var/lib/exim4/config.autogenerated.
> Is that the intended meaning?  My guess is the intended meaning is
> If update-exim4.conf detects that the freshly generated configuration
> file is invalid it exits with an error.  Unless the -o option was
> given, the old output file is retained.
Hej,

Afaict the documentation is correct, update-exim4.conf does not check
the validity if -o is given and does not exit with an error if -o is
given and the file is invalid:

---------------------
argenau:/etc/exim4/conf.d/main# echo blah > invalid
argenau:/etc/exim4/conf.d/main# update-exim4.conf
2006-05-21 10:54:00 Exim configuration error in line 201 of
/var/lib/exim4/config.autogenerated.tmp:
  main option "blah" unknown
Invalid new configfile /var/lib/exim4/config.autogenerated.tmp
not installing /var/lib/exim4/config.autogenerated.tmp to 
/var/lib/exim4/config.autogenerated
argenau:/etc/exim4/conf.d/main# update-exim4.conf -o /tmp/xxxxx
---------------------

           cu andreas
-- 
The ''Galactic Cleaning'' policy undertaken by Emperor Zhark is
a personal
vision of the emperor''s, and its inclusion in this work does not
constitute
tacit approval by the author or the publisher for any such projects,
howsoever undertaken.                                (c) Jasper Ffforde

On Sun, May 21, 2006 at 10:57:29AM +0200, Andreas Metzler
wrote:
> On 2006-05-20 Ross Boylan <ross@biostat.ucsf.edu> wrote:
> > The man page for update-exim4.conf says, in part
> 
1)
> >   If update-exim4.conf is called without the option -o, it
> >   checks the validity of the freshly generated configuration file and
> >   exits with  an error.   In   that   case,   the   new   file   is
> >   not  installed  to /var/lib/exim4/config.autogenerated.
> 
> > Is that the intended meaning?  My guess is the intended meaning is
> 
2)
> > If update-exim4.conf detects that the freshly generated configuration
> > file is invalid it exits with an error.  Unless the -o option was
> > given, the old output file is retained.
> 
> Hej,
> 
3)
> Afaict the documentation is correct, update-exim4.conf does not check
> the validity if -o is given and does not exit with an error if -o is
> given and the file is invalid:
I think 1, 2, and 3 above all mean different things.
First, with -o:
1) gives no information about what happens when -o is used.
2) says when -o is used there is a check, and output occurs in all
cases
3) with -o there is never an error check and output always occurs.
1) is consistent with 2) or 3), but implies neither.  2) and 3) are
inconsistent.  Since 2) is my guess, it''s probably wrong.

Second, without -o:
1) validity is checked and an error exit occurs.
My original concern was that 1) literally means an error exit *always*
occurs, even if the file is valid.  The new file is *never* installed
(since "in that case" refers to something that always happens).
2) validity is checked and an error exit occurs if the generated file
is invalid.  If there is an error, the old file is retained;
otherwise, the new one is installed.
3) gives no information on this case
> ---------------------
> argenau:/etc/exim4/conf.d/main# echo blah > invalid
> argenau:/etc/exim4/conf.d/main# update-exim4.conf
> 2006-05-21 10:54:00 Exim configuration error in line 201 of
/var/lib/exim4/config.autogenerated.tmp:
>   main option "blah" unknown
> Invalid new configfile /var/lib/exim4/config.autogenerated.tmp
> not installing /var/lib/exim4/config.autogenerated.tmp to 
> /var/lib/exim4/config.autogenerated
> argenau:/etc/exim4/conf.d/main# update-exim4.conf -o /tmp/xxxxx
> ---------------------
> 
These results are consistent with all 3 interpretations, though only
weakly consistent with 3), which doesn''t discuss the not -o case.

Ross

On 2006-05-21 Ross Boylan <ross@biostat.ucsf.edu>
wrote:
> On Sun, May 21, 2006 at 10:57:29AM +0200, Andreas Metzler wrote:
>> On 2006-05-20 Ross Boylan <ross@biostat.ucsf.edu> wrote:
>>> The man page for update-exim4.conf says, in part
> 1)
>>>   If update-exim4.conf is called without the option -o, it
>>>   checks the validity of the freshly generated configuration file
and
>>>   exits with  an error.   In   that   case,   the   new   file   is
>>>   not  installed  to /var/lib/exim4/config.autogenerated.
>>> Is that the intended meaning?  My guess is the intended meaning is
> 2)
>>> If update-exim4.conf detects that the freshly generated
configuration
>>> file is invalid it exits with an error.  Unless the -o option was
>>> given, the old output file is retained.
>> Hej,
> 3)
>> Afaict the documentation is correct, update-exim4.conf does not check
>> the validity if -o is given and does not exit with an error if -o is
>> given and the file is invalid:
[ snip]
> My original concern was that 1) literally means an error exit *always*
> occurs, even if the file is valid.  The new file is *never* installed
> (since "in that case" refers to something that always happens).
I can see the problem now.
> 2) validity is checked and an error exit occurs if the generated file
> is invalid.  If there is an error, the old file is retained;
> otherwise, the new one is installed.
> 3) gives no information on this case
[...]

How about:

(Only) if update-exim4.conf is called without the option -o, it
checks the validity of the freshly generated configuration file. If
the new file is invalid it is not installed to
/var/lib/exim4/config.autogenerated and update-exim4.conf exits with
an error. 
     cu and- no native speaker -reas
-- 
The ''Galactic Cleaning'' policy undertaken by Emperor Zhark is
a personal
vision of the emperor''s, and its inclusion in this work does not
constitute
tacit approval by the author or the publisher for any such projects,
howsoever undertaken.                                (c) Jasper Ffforde

On Sun, 2006-05-21 at 20:14 +0200, Andreas Metzler wrote:
> (Only) if update-exim4.conf is called without the option -o, it
> checks the validity of the freshly generated configuration file. If
> the new file is invalid it is not installed to
> /var/lib/exim4/config.autogenerated and update-exim4.conf exits with
> an error. 
>      cu and- no native speaker -reas
That''s pretty good.  How about

update-exim4.conf checks the validity of the freshly generated
configuration file if and only if invoked without the -o option.  If the
new file is invalid, update-exim4.conf exits with an error and leaves
the old /var/lib/exim4/config.autogenerated untouched.

It might be nice to check the validity of the output file in all cases
(which was the behavior I assumed), but that''s a somewhat separate
issue.

-- 
Ross Boylan                                      wk:  (415) 514-8146
185 Berry St #5700                               ross@biostat.ucsf.edu
Dept of Epidemiology and Biostatistics           fax: (415) 514-8150
University of California, San Francisco
San Francisco, CA 94107-1739                     hm:  (415) 550-1062

On Mon, May 22, 2006 at 09:47:56AM -0700, Ross Boylan
wrote:
> update-exim4.conf checks the validity of the freshly generated
> configuration file if and only if invoked without the -o option.  If the
> new file is invalid, update-exim4.conf exits with an error and leaves
> the old /var/lib/exim4/config.autogenerated untouched.
> 
> It might be nice to check the validity of the output file in all cases
> (which was the behavior I assumed), but that''s a somewhat separate
> issue.
So you want something that could be documented with:

update-exim4.conf checks the validity of the freshly generated
configuration file. If the new file is invalid, update-exim4.conf
leaves the old /var/lib/exim4/config.autogenerated untouched, but will
overwrite an output file specified with -o.

Right?

Genereally, the update-exim4.conf man page should be flagged for a
major re-work.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don''t trust Computers. They | Mailadresse
im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835

On Tue, May 23, 2006 at 05:51:37PM +0200, Marc Haber
wrote:
> On Mon, May 22, 2006 at 09:47:56AM -0700, Ross Boylan wrote:
> > update-exim4.conf checks the validity of the freshly generated
> > configuration file if and only if invoked without the -o option.  If
the
> > new file is invalid, update-exim4.conf exits with an error and leaves
> > the old /var/lib/exim4/config.autogenerated untouched.
I have committed the following new wording:

       If the new configuration will be written to
       /var/lib/exim4/config.autogenerated, update-exim4.conf will
       check the validity of the freshly generated configuration.  If
       the new file is detected as invalid, update-exim4.conf leaves
       the old /var/lib/exim4/config.autogenerated untouched and exits
       with an error.

       However, there are still possible invalidities that can only be
       detected at run time. This most notably applies to errors in
       expressions that are expanded at run time.

       If the new configuration will be written to some other file,
       this file''s contents will always be overwritten regardless of
       the validity of the freshly generated configuration.

> > It might be nice to check the validity of the output file in all cases
> > (which was the behavior I assumed), but that''s a somewhat
separate
> > issue.
I disagree on that point. The current behavior is just fine in my
opinion.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don''t trust Computers. They | Mailadresse
im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835

On Mon, 2006-06-26 at 14:26 +0200, Marc Haber wrote:
> On Tue, May 23, 2006 at 05:51:37PM +0200, Marc Haber wrote:
> > On Mon, May 22, 2006 at 09:47:56AM -0700, Ross Boylan wrote:
> > > update-exim4.conf checks the validity of the freshly generated
> > > configuration file if and only if invoked without the -o option. 
If the
> > > new file is invalid, update-exim4.conf exits with an error and
leaves
> > > the old /var/lib/exim4/config.autogenerated untouched.
> 
> I have committed the following new wording:
> 
>        If the new configuration will be written to
>        /var/lib/exim4/config.autogenerated, update-exim4.conf will
>        check the validity of the freshly generated configuration.  If
>        the new file is detected as invalid, update-exim4.conf leaves
>        the old /var/lib/exim4/config.autogenerated untouched and exits
>        with an error.
> 
>        However, there are still possible invalidities that can only be
>        detected at run time. This most notably applies to errors in
>        expressions that are expanded at run time.
> 
>        If the new configuration will be written to some other file,
>        this file''s contents will always be overwritten regardless
of
>        the validity of the freshly generated configuration.
> 
> 
> > > It might be nice to check the validity of the output file in all
cases
> > > (which was the behavior I assumed), but that''s a
somewhat separate
> > > issue.
> 
> I disagree on that point. The current behavior is just fine in my
> opinion.
It might be good to make this behavior a bit more explicit.  Here''s a
possible rewording of the last paragraph:
       If the new configuration will be written to some other file,
       no validity checking occurs  and that file will always be
       overwritten.
Ross

On Tue, Jun 27, 2006 at 01:48:22PM -0700, Ross Boylan
wrote:
> It might be good to make this behavior a bit more explicit. 
Here''s a
> possible rewording of the last paragraph:
>        If the new configuration will be written to some other file,
>        no validity checking occurs  and that file will always be
>        overwritten.
Very nice. I have committed this wording to svn.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don''t trust Computers. They | Mailadresse
im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835