newscache: Thread: Wildcard lists

Hi everybody.
I use NewsCache since the 0.99.17 release. I have recently upgraded to the
newest releases. I tested both stable and development versions but I still
have a problem with wildcards in newscache.conf.
In my conf file I have this server configuration:

Server news.xxx.yy nntp {
    Retries 3
    Read *,!it.*,!alt.*
    PostTo *,!it.*,!alt.*
    GroupTimeout 150
    Options cached
    Commands not-over
 }

I want to fetch from this server all newsgroups, without it.* and alt.*
hierarchies but in my active list I find all newsgroups, including it.* and
alt.*.

Using this line:

Read *,!it.*

it works fine, I get all groups, except it.* but if I have more than one
exclusion in wildcard line it works like a Read *

The multiple exclusion in the very old NewsCache 0.99.17 works fine.

Anybody has a hint?

Bye

Marco Restani

####@####.#### wrote:

>Hi everybody.
>I use NewsCache since the 0.99.17 release. I have recently upgraded to the
>newest releases. I tested both stable and development versions but I still
>have a problem with wildcards in newscache.conf.
>In my conf file I have this server configuration:
>
>Server news.xxx.yy nntp {
>    Retries 3
>    Read *,!it.*,!alt.*
>    PostTo *,!it.*,!alt.*
>    GroupTimeout 150
>    Options cached
>    Commands not-over
> }
>
>I want to fetch from this server all newsgroups, without it.* and alt.*
>hierarchies but in my active list I find all newsgroups, including it.* and
>alt.*.
>
The man page newscache.conf doesn't describe the real situation with the 
Server configuration parameters. I found this at the same time as 
Konstantin Nikonenko sends his regular expression patch one month ago. I 
think, this can't work, because the NewsCache send a list active command 
to his upstream NewsServer (he split the Read Parameter by "," and sends 
one command per part. Example:

list active *
list active !it.*
list active !alt.*

You can find this in the /var/log/news/news.debug file (other example):
Jan 15 11:32:31 test NewsCache[18721]: CServer::active: active database 
timed out
Jan 15 11:32:31 test NewsCache[18721]: RServer::active()
Jan 15 11:32:31 test NewsCache[18721]: 
RServer::setserver({news.wienkav.at,nntp})
Jan 15 11:32:31 test NewsCache[18721]: RServer::connect()
Jan 15 11:32:31 test NewsCache[18721]: RServer::connect: Connecting to 
up.stream.newsserver from DEFAULT inte
rface to servicename nntp
Jan 15 11:32:31 test NewsCache[18721]: sstream::connectTo connecting to 
up.stream.newsserver service nntp
Jan 15 11:32:31 test NewsCache[18721]: 200 NewsCache 1.1.91, accepting 
NNRP commands
Jan 15 11:32:31 test NewsCache[18721]: RServer::issue: issue mode reader
Jan 15 11:32:31 test NewsCache[18721]: RServer::issue: issue list 
overview.fmt
Jan 15 11:32:31 test NewsCache[18721]: RServer::issue: issue list active *
Jan 15 11:32:31 test NewsCache[18721]: NVActiveDB::read(&is,*filter,flags)
Jan 15 11:32:31 test NewsCache[18721]: RServer::issue: issue list active 
!it.*
Jan 15 11:32:31 test NewsCache[18721]: NVActiveDB::read(&is,*filter,flags)

OK, and try this commands in a telnet session! In the version 1.1.92 the 
man page contains for this parameter:

       Read NewsgroupList
              The  comma separated list specifiy the wildmat newsgroup 
specifications. Each element of
              the list will be fetched with the list active element. 
Example: Read at.*,de.* will send
              to  the upstream newsserver the following two commands: 
list active at.* and list active
              de.*. For details see section 2.1.2 LIST ACTIVE [WILDMAT]. 
These parameter can help  you
              saving bandwidth.  AccessSpec Configuration Options
               . No default value.

And now zless ../doc/rfc2980.txt.gz

2.1.2 LIST ACTIVE

   LIST ACTIVE [wildmat]

   LIST ACTIVE is exactly the same as the LIST command specified in RFC
   977.  The responses and the format should exactly match the LIST
   command without arguments.  If the optional matching parameter is
   specified, the list is limited to only the groups that match the
   pattern.  Specifying a single group is usually very efficient for the
   server, and multiple groups may be specified by using wildmat
   patterns (described later in this document), not regular expressions.
   If nothing is matched an empty list is returned, not an error.  This
   command first appeared in the UNIX reference version.

and

3.3 The WILDMAT format

   The WILDMAT format was first developed by Rich Salz based on the
   format used in the UNIX "find" command to articulate file names.  It
   was developed to provide a uniform mechanism for matching patterns in
   the same manner that the UNIX shell matches filenames.  Patterns are
   implicitly anchored at the beginning and end of each string when
   testing for a match.  There are five pattern matching operations
   other than a strict one-to-one match between the pattern and the
   source to be checked for a match.  The first is an asterisk (*) to
   match any sequence of zero or more characters.  The second is a
   question mark (?) to match any single character.  The third specifies
   a specific set of characters.  The set is specified as a list of
   characters, or as a range of characters where the beginning and end
   of the range are separated by a minus (or dash) character, or as any
   combination of lists and ranges.  The dash can also be included in
   the set as a character it if is the beginning or end of the set.
   This set is enclosed in square brackets.  The close square bracket
   (]) may be used in a set if it is the first character in the set.
   The fourth operation is the same as the logical not of the third
   operation and is specified the same way as the third with the
   addition of a caret character (^) at the beginning of the test string
   just inside the open square bracket.  The final operation uses the
   backslash character to invalidate the special meaning of the a open
   square bracket ([), the asterisk, backslash or the question mark.
   Two backslashes in sequence will result in the evaluation of the
   backslash as a character with no special meaning.

3.3.1 Examples


   a. [^]-] -- matches any single character other than a close square
               bracket or a minus sign/dash.

   b. *bdc  -- matches any string that ends with the string "bdc"
               including the string "bdc" (without quotes).

   c. [0-9a-zA-Z] -- matches any single printable alphanumeric ASCII
               character.

   d. a??d  --  matches any four character string which begins
                with a and ends with d.

If you would wipe out the newsgroup lists for a user, then you have to 
specify this in the AccessSpec section. And in the AccessList section 
the regular expression patch from Konstatin is very usefull, but we need 
more development work to do this. The Server Read option is usefull, if 
you combine multiple upstream server. Here you can specifiy, which 
Newsgroup is taken from a specific server. But attention: The PostTo 
works only with the primary server at the moment: man newscache.conf 
(1.1.92):

      PostTo NGSpecification
              This function is not implemented at the moment. NewsCache 
post an article to the primary
              news server. The primary news server is the news server 
being listed first. If the arti-
              cle cannot be submitted immediately, it is kept for later 
transmission.

Any suggestions?


>
>Using this line:
>
>Read *,!it.*
>
>it works fine, I get all groups, except it.* but if I have more than one
>exclusion in wildcard line it works like a Read *
>
>The multiple exclusion in the very old NewsCache 0.99.17 works f
>

Are you sure?