does your critique also apply to the previous message in this thread whose author suggested using anubis?
On Fri 6. Mar 2026 at 13:14, Crystal Kolipe <[email protected]> wrote: > On Fri, Mar 06, 2026 at 12:17:44PM +0200, ????????? wrote: > > A bit off topic, but what about putting the ole CVSWEB behind some form > of > > anonymous registration > > Why? > > What problem would this solve? > > If there are bugs or missing functionality in the new cvsweb, then the > logical > thing to do would be to fix them there. > > If you're suggesting to run both old and new in parallel, have you > considered > the additional admin burden that would put on Nick? > > > (maybe even send your self signed client certificate > > per mail), and refuse all who are not registered? > > So then you have to deal with 'bug reports' flooding in from people who let > their certs expire and didn't realise it, or just don't know how to > configure > them propperly in the browser. > > And from a practical point of view, the client certificate idea makes > little > sense - one of the uses of cvsweb is that it's a way of looking at the > codebase from _any_ convenient web browser. If you've got to make sure > that > your personal client cert is installed in advance, the utility of the > service > drops considerably. > > > That would be a small > > wall which robots can't easily climb. > > The original problem was the load on the server from badly designed bots, > not > outright prevention of bots accessing the codebase. > > The most correct solution to an overloaded server is either to make it more > efficient, (for which there was plenty of scope to do that in this case), > or > to upgrade the server hardware, (which has a direct monetary cost and also > the > risk of downtime and maintenance burden due to unexpected hardware issues, > etc, etc). > > Trying to fix an overloaded server by restricting the number of users who > can > connect to it is a rubbish solution in general. Maybe as a temporary fix, > but > long term it's not a professional approach. >
