0.79 releases

[lurker]

Is it possible for the number of teams to be less than the number of ally teams? If so, has a game ever been run that way? The only situation I can see this syncing on .1 but desyncing with .2 is if you have a 1v1 or FFA and the ally teams don't line up with the teams. This is probably uncommon enough to not care about. It technically breaks sync, but you practically have to force it to.

[hoijui]

i dont get it...
how could a 1vs1 have 3 allyTeams?

(engine internally, the team and allyTeam numbers get reassigned, so they all start at 0 and aend at n-1)

[lurker]

I don't understand why you're asking that question, but Gaia.

I'll restate the two ways I see to desync. Way one is adding a bunch of extra ally teams with no players. This may or may not work. Way two is to have team 0 be on ally 1, and team 1 on ally 0. This seems like it would be rare.

[Licho]

I dont understand what is lurker talking about ..
but ..
people with version other than 0.2 DO desync and DO crash within couple of minutes into game.

[hoijui]

mmm ... i may still not get what you mean, lurker, but if i get it, then there is no possible problem there.

[AF]

Maybe we should talk about what to do next rather than discuss whats already been done?

[hoijui]

or maybe, we are already there?

.. because the path is the goal?

[lurker]

I'm talking about a very narrow situation where:
79.0.1 players sync with each other
79.0.2 players sync with each other
79.0.1 players don't sync with 79.0.2 players

I find it unlikely that this has happened in any real games. Why am I talking about this situation? To disagree with Kloot saying that it's a version violation because it "breaks sync with 0.79.0.1"

Most of the behaviour of this bug is undefined on 79.0.1 and as such there is no issue of breaking sync, no sync to break.

[el_matarife]

If we're really close to a .79.0.3 release with a few more bug and crash fixes should you really bother forcing people to upgrade? There's at least two crash bugs that have been fixed since .79.0.2, maybe more, and Spring is updated to the latest version of LUA. I assume there's going to be another release to get all these changes out to the general public, but I don't know if this is in the next week or in the next month.

[imbaczek]

we're closer to 79.1, actually. it'll be a forced upgrade. no ETAs but expect a rc soonish.

[Tobi]

Current master is fine to become 0.79.1.0 IMO (nearly only bugfixes have been committed since 0.79.0.0)

I'd have uploaded a RC now if only it did compile on BuildServ. No clue how to fix the error it gives... (even !rebuild clean=yes kept giving error, while my fix did make it work for baczek)

[imbaczek]

figured out why i thought it did work - it didn't, it just didn't manage to compile the ai before barfing on something else which i broke and didn't bother fixing.

anyway, the culprit is in the generated java (OOAI, AbstractOOAI):

Code: Select all

	public int unitCreated(Unit unit, int builder) {

[Kloot]

Ah, my bad, my setup only compiles native AI's so I didn't notice that (and my grep string was 'UnitCreated' ).

lurker: I believe I already acknowledged the rarity of the situation.

[Auswaschbar]

I'd say end of the week for 0.79.1 (Friday, probably).

[Tobi]

More pending (synced) bugfixes?

[Licho]

I dont understand what is lurker talking about ..
but ..
people with version other than 0.2 DO desync and DO crash within couple of minutes into game.

Unlikely? Wtf.. its every bigger game for me! Every bigger game i see someone with 0.1 joining, i warn him, he still plays and desyncs or crashes..

I can send you tens of demos of this just from few days...

[jK]

More pending (synced) bugfixes?

Yeah, I want to change the arguments of Spring.SetUnitShieldState, but I am very busy atm (and can't merge the current repo).
I hope to find a bit time until Friday to commit it.

[Jools]

I don't understand what the problem is, what's so difficult with telling people to use the 0.79.0.2 version? It's just one exe to replace.

[imbaczek]

the problem is, they don't listen.

[Tobi]

More pending (synced) bugfixes?

Yeah, I want to change the arguments of Spring.SetUnitShieldState, but I am very busy atm (and can't merge the current repo).
I hope to find a bit time until Friday to commit it.

What do you want to change it to and do you mind if I just apply this change?

(I looked at it yesterday and as it is now with Kloots fix it seems fine; behavior in 0.79.0.* is totally broken cause 'enabled' and 'weaponIndex' is the same argument number [ => enabled can not be set at all in 0.79.0.* ], so keeping compat with this is pointless anyway.)