Meh=-~ wrote:idc about my grammar and stuff on a dumb ass forum
Get rid of that attitude, please. Quickly. Writing properly is a sign that you respect your readers, consequently not writing properly is a sign of your disrespect. Why would anyone listen to you if you show how little you care?
On the issue, it seems like the client failsafe system isn't working properly. It only tries ONE other master instead of all of them. Workaround: leave the server browser and try again. For about 60 seconds, non-working masters are not polled again. 'Course, I'll do a proper fix for 0.2.8.3.
On the proper datacenter hosting: I think tank's servers are running that way already. It doesn't really help

I know that I'd notice a total outage of my home server hosting my master much sooner than I'd notice the outage of my rented VPS. And no, there's no hardcoded master count limit.
Regarding the slowness, yeah, it's the many servers. There's also the fact that they all get pinged in arbitrary order. I have the code here for 0.2.8 that can make servers that had a higher score last time get pinged first, I just need to extract it from some unrelated change. I think it would help make the popular servers visible more quickly (and if you like an unpopular server, you can always give it a score bias) and bury "Unnamed Server 15". Also, we're already thinking about a better browser structure for trunk versions. A flat list was fine for 20 servers, but not so any more.