Installation
    ------------

    Please read the file INSTALL!

    Here is a quick summary:

    1)	Unpack the TAR.  With GNU tar you can do it this way:
	"tar xzvf gulp.tar.gz"

    2) "cd gulp"

    3)	"./configure"

    4) "make"

    5) "make install"

    6) You end up with a "gulp" binary in /usr/local/bin.


    Usage
    -----

    Gulp is a program for transporting news articles between news
    servers using NNTP.  When invoked with a source and a destination
    host it will look for new articles in each group in the
    destination server's active file on the source host.  Once a list
    of articles has been compiled it will fork a number of times and
    each child will transfer a proportion of the required articles in
    parallel.

    Optionally a the name of a file containing a perl routine
    filter_xover() may be specified.  This is passed the name of the
    newsgroup and then the entries in the overview database for the
    article in question. The script should use that to decide whether
    the article should be accepted or not, returning the scalar OK if
    it is to be accepted and anything else for a rejection (the
    routine's return value is then reported as the reason for
    rejection).

    If the source newsserver will allow pipelining of article commands
    then that should be turned on (by adding '-p' to the command line)
    for extra throughput.  It is possible that if the source
    newsserver also requires authentication then it isn't possible to
    take advantage of pipelining since any ARTICLE command could be
    replied to with '480 authentication required' and the program
    can't respond to that since it has already queued a number of
    other article requests.  You may be able to get round this by
    forcing authentication when the client initially connects (by
    using the '-a' option) but this is not guaranteed to work and you
    may just have to turn pipelining off.

    Example
    -------

    To transfer articles from news.demon.co.uk to meow.netscum.org.uk
    using 6 parallel streams of articles and a killfile use the
    following command line:

        gulp -F 6 -k filter.pl -f news.demon.co.uk -t meow.netscum.org.uk\ 
		-n newsrc -A active.ignore

    The file newsrc is just used to store state information between
    runs and active ignore is a list of newsgroups present in
    meow.netscum.org.uk's active file which shouldn't have new
    articles fetched for them.

    The file filter.pl might look like this

             sub filter_xover {
                 my ($newsgroup,$subject,$author,$date,$messageid,
                     $references,$bytes,$lines,$xref) = @_;
                 if($bytes > 100000) {
                   return ("Exceeds 100k");
                 }
                 return("OK");
              }

    if you wanted to avoid transferring any article larger than 100
    kilobytes.  The various fields of the overview data and the
    newsgroup name will always be transfered as a list of scalars as
    indicated.