I suppose I really ought to bring this up again, so we can finish the issue. *sigh*.
Raul continues to suggest using a CGI program: > The cgi script wouldn't have to run on the same system. It's > a pretty simple script (see below), just change input mechanism > from <> to something based on use CGI; Yes, it is simple. But, regardless of whether you run the CGI script on the same system you still concentrate the load and points of failure. We have more mirrors than places we can run CGI scripts. Hopefully we'll eventually have a Debian package (in Extra) that maintains a local bug mirror. The current arrangement with the CGI script for going straight to a bug by number or a package already has lower availability than reports themselves. I think that we really have only the following 3 options: (a) Leave things as they are. (b) Reverse the order into forwards-chronological. (c) Supply both sets of pages. If we have to choose from (a) and (b) I think I'm convinced that we'll have to go for (b), so no more flameage about that please. The question then is whether (c) can be made to be acceptable. Do we need a complete set of indexes with links to the reversed pages, or two hyperlinks to each bug on each index ? Or can we just put a link at the top of the forward version into the reversed one ? Is the extra data storage a problem ? Ian. -- TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to [EMAIL PROTECTED] . Trouble? e-mail to [EMAIL PROTECTED] .