Devblog #7 (19th June 2016)
Demo still not released. Not surprised in the slightest. Every week that I say I’m planning to release the demo the following week, some new bug or something crops up and prevents me from doing so. Therefore, this is my official announcement that I have no hopes of releasing the demo this week.
Ok, here’s a breakdown of the demo delays this week:
Uploaded dedicated server binaries to steam. On testing, they seemed to work when entering the server I.P manually, but didn’t appear on the server list. Was informed that the Steamworks documentation for dedicated servers is currently out of date, and would be updated on Tuesday. Wasn’t a great deal I could do without accurate documentation, so I waited till Tuesday.
On Tuesday, while waiting for the documentation to be updated, I did some more testing with 3 steam users, using a listen server. This revealed some more bugs, the most serious of which kicked a player to the menu seemingly at random. I fixed this the next day (so far as I can tell at least).
Here is a description of the bug, and the process of fixing it, if that kind of thing floats your boat:
What I believe the problem was was that some kind of buffer for networked functions in UE4 was being overloaded. Presumably this is some kind of safety feature to stop a client overloading everyone’s connections. Anyway, I worked out (from the game logs) that the function causing the problem was the one that reports a clients current aim direction to the server. I reduced the bandwidth used by the function by using compressed data, and made sure the function was called less regularly by only calling it if the client’s aim had changed (Should have done that already, really, but this particular function has probably been in the game for over a year, so I’d forgotten about it) and also, setting a maximum call rate of 120Hz (since it later transpired that the player who encountered the bug had an enormous framerate of over 300fps, which would have helped in triggering the bug).
Anyway, having spent Wednesday fixing this new bug, it seemed to be fixed upon testing. Although some more bugs were encountered in this testing process, which I then spent much of Thursday fixing. They weren’t very interesting. Made a few other minor improvements while I was at it. On Friday, I finally got round to looking at Valves new documentation for dedicated servers on steam. Followed the instructions for allowing the server to run anonymously (I.E not being associated with a specific steam account), which I think might have something to do with why they don’t currently show up on the server list. Having followed the instructions though, it doesn’t seem to have worked. Am still waiting for a response about this.
I was also meaning to fix a few minor issues with the Linux build this week, but wasn’t expecting so much trouble with the servers.