README.website-split

As of October, 2005, the Fink website has three portions, hosted in different
places.  This CVS repository contains all files used by all portions.

Thus, to fully update the website requires three steps:
 1) update the content in CVS;
 2) log on to ssh.sourceforge.net, change to the fink htdocs directory,
    and run ./update.sh ;
 3) after an appropriate delay, log on to sampson.opendarwin.org, cd to
    /Volumes/src2/fink/finkbindist , and run ./update.sh ; then cd to
    /Volumes/src2/fink/pdb.finkproject.org/htdocs and run ./update.sh .
(The reason for a delay in step 3 is that the CVS checkout of the web
content is via anonymous pserver access, which is only updated at sourceforge
after a delay.)

Since the bulk of the website remains at sourceforge.net, for most updates
step 3 is not necessary.

The files download/nav.*.inc contain explicit pointers to 
bindist.finkmirrors.net/bindist/dists/ (which is hosted at 
sampson.opendarwin.org), which is how the split works for users.  On the
other hand, bindist/index.php sets up the index bar on the left of the page
to point back to the sourceforge.net site.

Similarly, the file header.inc redirects the 'packages' part of the website
to pdb.finkproject.org .  The file pdb/header.inc makes sure that any
stray URL's of the form fink.sourceforge.net/pdb/... get redirected to the
pdb.finkproject.org site.

The .cvsignore file contains an entry for index.shtml; on the two remote
sites, this is an empty file which prevents the web content one level
above that of the remote site from being displayed by the remote webserver
(while allowing us to use the same CVS files for both the main and the
remote sites).