Back after 3 months, what have I missed?

Last online: September 5th, 2014
So, after 3 months of not opening spring at all, and then 30 minutes of downloading chatlogs for #zk and #zkdev (enough to make Weblobby take ~565MB), I see that not much has changed. Spring 98.0 has come out, but for some absurd reason we're still using 91.0 for the default matches. Please, for the love of God, tell me that you've at least fixed the desync errors in 91.0; otherwise we still don't really have any version that works 100% of the time for everyone. I still remember insisting that 97.0 runs as fast as 91.0 on my computer, and although the lag bugs didn't seem to show up for everyone, iirc Licho was having lag issues for 97.0, so you should have at least had someone available to lag testing without complaining.
I've also noticed that neither the steam group page, nor the steam greenlight page for ZK have been touched by any name that I vaguely recognize since august. The most recent poll was on this topic, and had results similar to what I'd expect, but I don't see why any change to ZKLobby would need to restrict it to steam without any standalone version.
Also, I realize I should have properly reported this bug 3 months ago, but when running ZKLobby in mono on linux I am unable to scroll down all the way in the players list in #zk and other rooms, the last names are below the furthest I can scroll, and I can see one that is partly cut off most of the time. This problem becomes worse when the size of the window is increased for some reason, i.e. I can scroll down to names starting with U when windowed, but only down to names starting with T when the window is fullscreened.

So, I'd like to ask what bugs in 98.0 are currently blocking you from updating, even after all this time; And if these bugs are still valid, is 91.0 actually playable now? (I personally don't count desyncs 5-10% of the time as "playable")

quote: The most recent poll was on this topic, and had results similar to what I'd expect, but I don't see why any change to ZKLobby would need to restrict it to steam without any standalone version.

To elaborate, Licho is adding proper quickmatching queues (rather than the clumsier room juggling system of before). However, because Steam has an API that handles matchmaking properly, there was some discussion about whether it would be better to just use the Steam API and not bother with Uberserver anymore. As expected, that isn't happening, and the queues being implemented as dummy rooms is still the direction things seem to be headed (for compatibility with other lobbies).

quote: then 30 minutes of downloading chatlogs for #zk and #zkdev

Please tell me how to do that!

quote: I still remember insisting that 97.0 runs as fast as 91.0 on my computer

There appears to be a trend between the brand of hardware (i.e. Intel vs AMD) versus the degree of improvement or reduction when using 98.

quote: How are the desyncs now? Are they still happening daily?

Yes, but usually on specific maps, independant of playercount.
Yes, but only if specific players are in the game; it is always the same guys who desync first, independant of playercount.
The rate of desyncs with neither of those involved is well below 5%.

It's a shame that ZK does not locally permanently store pertinent information when desyncs occur, or automatically upload that information to some central site to allow for tracking of which players and what maps experience the most desyncs.

Is such information storage and uploading against some ethical principle?

quote: Yes, but usually on specific maps, independant of playercount.

Playercount does seem to have an effect. Not sure if directly or only indirectly. I've probably played over 800 duels and I recall 0 desyncs in all of them. There might have been one a very long time ago, but I'm not 100% sure it was a duel. Meanwhile some who play large games have recently been complaining about desyncs quite often.

quote: I've probably played over 800 duels and I recall 0 desyncs in all of them.

It happens sometimes. The last one I can remember was months ago (8? 12?). I vaguely remember that it was me vs klon on avalanche using 91.0.

quote: Playercount does seem to have an effect. Not sure if directly or only indirectly. I've probably played over 800 duels and I recall 0 desyncs in all of them. There might have been one a very long time ago, but I'm not 100% sure it was a duel. Meanwhile some who play large games have recently been complaining about desyncs quite often.

Assuming it's individual players causing/having desync, then of course player count scales the amounts of desync occuring.
Keep in mind that 10x as many players not only means 10x as high of a chance for a desync, but also 10x as many players being affected by it.

In other words, the chance to witness a desync grows quadratically with game size.

arrrrggg! what have you done? now everyone gets quadratical!

why? couldn't you just fix the issue causing the desync? or... hell when the person does desync set their army to an AI for a bit until they have synced back up?

AFAIK it's impossible to correct simulation after desyncing. It only gets worse with time.

@Anarchblah1, yes.

Benchmark isn't really needed, we played a 3v3 testgame on CCR and all or most players had performance issues and other bugs. The biggest problem is probably LoS map turning black on certain maps for some people on F1/F2, or becoming brighter and brighter until it's white for other people. Second biggest problem for me is units randomly deciding to move into the top left corner and other pathing issues. Third are FPS drops.