Spring 0.82.6 bug fix release

[hoijui]

Thanks for the testing and bug fixing!
This release fixes some of the most prominent path-finding issues, the FPS crash bug and the unitsync map-name issues, among other stuff.

Download links:

• Windows installer
• Windows portable archive - Alternatively, use the Portable option in the windows installer
• Source (tar.lzma)
• Source (tar.gz)

Detailed changes since 0.82.5.1:

Engine / General:

• fix various bugs related to mid-game join
• fix LosHandler ignoring units with losRadius <= 0 but airLosRadius > 0
• fix a leftover icon bug
• fix BFGroundDrawer crash at extreme view-distances
• fix death-wait
• allow death-wait in games with more then 2 ally teams
• reduce risk for deadlock during stack trace
• allow to specify game-start-delay in the start script (GameStartDelay)

Engine / Simulation:

• prevent SEGV when updating skidding units
• improve turnInPlace=0 path-following (#2072)
• units ignored ETA failures when <= 200 elmos from goal, even if goal unreachable
• make pushResistant (allied) units still move out of the way of constructions

Engine / Rendering:

• update the displayed resource production for inbuilt units too
• fix spectators see ghosted buildings
• do not draw icons for noDraw units
• do not cast shadows for noDraw units
• do not draw healthbars for noDraw units either

Engine / AI:

• add feature-ID trace-ray command
• allow fetching of other teams resources (current, income, usage, storage)
• EnableCheatEvents() was a no-op because it did not enable cheats temporarily
• IsUnitInLosOrRadarOfAllyTeam() now considers gs->globalLOS
• E323AI: new version: 3.22.4
• Python AI Interface: removed a few memory leaks

Unitsync:

• fix mapname/mapfilename conflicts

Installer:

• add portable install option (default: disabled)
• add Python AI Interface to the windows installer

Repository:

• remove SelectionEditor

[Masure]

You meant "you still need 0.82.5.1 to play online"

[hoijui]

no, it is right as it is.
0.82.5 includes 0.82.5.1 (which is just a patch release).
though i agree that we should change this naming policy. what should be changed:
all releases should have a 4 part version number, so there would not be a 0.82.5 release, but 0.82.5.0. this would remove confusions (like eg. this one) and be more consistent in general.

[Beherith]

Woohoo!

Thanks for the hard work to everyone who contributed!

[zwzsg]

You forgot to pack up MSVCR71.dll in the windows installer.

[yourface]

You forgot to pack up MSVCR71.dll in the windows installer.

So I'm not the only one...

[Gota]

yay,fixed pathing.good job.

[aegis]

no, it is right as it is.

sorry, I fixed it and didn't say anything

[zerver]

You forgot to pack up MSVCR71.dll in the windows installer.

So I'm not the only one...

Hmm, we shouldn't need that, static linking ftw.

[hoijui]

my findings...

Apparently, as spring.exe is now, it requires msvcr71.dll to run. This is a library implementing standard C functions like printf(). It is not guaranteed to exist in every windows. On WinXP for example, it does not come with a clean install (plus service packs and updates). msvcrt.dll is guaranteed to exist on all windows installations, and implements about the same functions, if i got it right (but... real below).

MinGW compiler comes with these libs:

• libmsvcr71.a
• libmsvcr71d.a
• libmsvcrt.a
• libmsvcrtd.a

MS has to say this:
http://msdn.microsoft.com/en-us/library ... S.71).aspx
(see "What is the difference between msvcrt.dll and msvcr71.dll?" on that page)

We shipped msvcr71.dll with the spring installer on the old build-bot (until 0.81). I will add it to the installer again.

[very_bad_soldier]

thanks guys...

[hoijui]

• added MSVCR71.dll to installer
• renamed SD
• changed hangTimeout from 10ms to 10s

thats on the release branch now, and will be on 0.82.6.1.
i think i will do that this evening or tomorrow, if no additions are required.

[Enjinn]

Is there any way of seeing advanced notification of the Spring engine change being required to play live? I appreciate this 0.82.6 isn't required for live yet, and hopefully not before tonight either. But, it would be very useful to old gimers (like a gamer but less so), like me, who only play once a week. I could move the date I play to work around the typical lead times I suffer getting the linux source re-compiled or wait for the PPA to update.

Much appreciated and keep up the outstanding work. BA is a time machine to happy days, mun.

[hoijui]

that is not possible, cause nobody without jedi powers knows that in advance.

[very_bad_soldier]

Lets Tobi do it then.

[Enjinn]

that is not possible, cause nobody without jedi powers knows that in advance.

Fine, everyone playing Imperial Winter knows in advance but what about the rest of us?

Seriously, is it a case of as the patch becomes ready someone sets it live? Just a shame there's no warning at all. Could there be a delay in getting the code out or is that not workable?

[smoth]

there already has been a delay.

[aegis]

this thread is the advance warning.

[Godde]

Pathing has improved in some areas.
But this is what happens S44 now:
http://springrts.com/phpbb/download/file.php?id=5071

It worked in 82.5
http://springrts.com/phpbb/download/file.php?id=5072

I tried changing turnInPlace=1 or even removing it but it didn't improve the behavior alot.

More pathing on http://springrts.com/phpbb/viewtopic.ph ... 48#p453148

Edit:
It is propably this fix "Mantis bugtracker:http://springrts.com/mantis/view.php?id=2072" that causes units with high maxVelocity and low turnRate to drive into walls and other obstacles.
When I raised turnRate or decreased maxVelocity the units pathed correctly.

[Argh]

The Java Null AI is the only Null AI included, and it isn't working properly.

It spams constant messages to Infolog, instead of, y'know, doing absolutely nothing, like it should be.

Either include a true Null AI, or allow the engine to be started without an AI at all that isn't the buggy, problematic Commanders script.

Also, reverse movement seems to be badly broken.