i think this is only possible if the client sets the referer correct, isnt it? nice idea btw. on the other hand, why cgi? we are using modules since quite a while now in-house, and it worked out to be pretty fast. the most used module is php/fi which makes perl based cgi scripts obsolete.
brian, how about the idea with splitting the document servers onto dns round robins with country pre/postfixes? like www.eu.apache.org, www.us.apache.org,... then a central server processor.apache.org which has a) some customized module for handling all that stuff that needs to be generated or b) some apache 1.2 with php/fi (thats what i use here) i dont know, but would it make sense to port bugdb.cgi to a serverside inline language? any ideas? btw, does the core team use cvsup? i already mentioned this before, it would be a nice way of having things mirrored. take care, -karsten On Fri, 13 Jun 1997, Hank Hughes wrote: > > Yikes, holy mail wars! > > How about the PERL/CPAN methodology? > > The mirror sites point all pivotal sections to *.cgi scripts > back to the mother ship: apache.org. then send "appropriate" > info back to the requestor looking for the closer node. > > For instance, list.cgi does not have to be mirrored but the > documentation/src that are mirrored should point to apache.org's > list.cgi. I'm probably not making any sense without sleep. So, > take a look at the way PERL/CPAN handles it's mirror: > > > ftp.funet.fi/pub/languages/perl/CPAN/ > > and our mirror which contians links that are requests for > "directions" from it's home site. > > ftp.ccs.neu.edu/net/mirrors/ftp.funet.fi/pub/languages/perl/CPAN/ > > > Sleeep, > -Hank > With best regards, Karsten W. Rohrbach -- Nuclear war can ruin your whole compile. (Karl Lehenbauer) -> http://www.webmonster.de -> http://www.nacamar.de -> http://www.quakeforum.de -> http://www.apache.de
