PPSSPP 0.9.9 release - blocking issues

[hrydgard]

Yes, it's time for a release again, I'm aiming for mid June. Last one before I'll have to declare 1.0 (well, we could name the next one 0.9.10 or 0.10 but I don't like that style of naming very much, and it risks confusion).

In this issue, let's only collect bad regressions from earlier versions. No feature requests.

• Innocent Life textures flickering and incorrect light (vertex cache) #6066 lighting issue
• Tilt controls are not working Qt #4907 (fixed)
• Problems with video intro BLEACH HEAT OF SOUL 6  #5535 (fixed)
• Graphical Regression in Dissidia 012 on Adreno 305 with PPSSPP build 695. #6107 (fixed)
• Rebuild font atlas

Also, buildbot must be fixed for a while.

[dbz400]

Are we consider the clut and block transfer branch for 0.99?

[daniel229]

flickering texture and incorrect light with hardware transform off in Innocent Life,since 9b1302c
it's a part of #6066

master build

reverted 9b1302c

[hrydgard]

@raven02 , if they are ready by end of May then maybe, otherwise I'm targeting them for 1.0.

[hrydgard]

@daniel229, hm, interesting little issue, I missed that in the big one. Will have to have a look.

[sum2012]

Ikki Tousen Xross Impact save broken between v0.9.8.1-46
#6068

[sum2012]

I forget important one,must repair Android build
#6069

[daniel229]

@sum2012 that hardware transform-error failing back to software issue,you post on the forum
http://forums.ppsspp.org/showthread.php?tid=11496

[sum2012]

@daniel229 Yes

[bhavin192]

#4907 (Tilt controls in symbian)
I know that a problem for symbian specific is not a blocker but still it's not working :(

@xsacha if you can please check, if you are well enough

[dbz400]

I think clut branch probably need few more work while block transfer should be good enough to see observable benefits with almost no regression

[solarmystic]

Time for 0.9.9 already? Felt like there wasn't much new between 0.9.8 and now, functionality wise, but it'd be great for users sticking with stable releases to get all the minor/major fixes introduced since then.

[hrydgard]

@solarmystic, yeah, I want to push out releases every 2 months or so, so that stable users aren't left too far behind. Last release was March 22 so it's soon time.

[xsacha]

I'll run some tests for Symbian to find out where it broke. Should be fixed by today.

[ghost]

#5192 god eater 2 mission error fix please :)

[thedax]

That's not a regression, though. That game has never worked properly (as far as I know) without a source code hack. Let's keep the discussion to regressions only.

[akemin-dayo]

Not exactly an emulation regression, but I submitted pull request #6083 that fixes a regression that causes PPSSPP to instantly crash on iOS 5 (and take SpringBoard with it) due to corruption of the autorelease pool.

[thedax]

I'm sure that's quite acceptable to discuss here, but I didn't author the thread. :P

[xsacha]

OK, so accelerometer just needs to be sent as axis events. I'll fix that for Blackberry and Symbian within the hour. Any other Symbian related blocker?

[thedax]

I forget important one,must repair Android build

It's been compiling fine manually for days, what happened to the buildbot?

[xsacha]

I can add Android to my buildbot if needed.

Edit: no need :)

[akemin-dayo]

http://ppsspp.angelxwind.net/builds/Android-fat/karenbuildbot-net.angelxwind.ppsspp-testing-android-fat.apk

http://ppsspp.angelxwind.net/?page/downloads#android

My buildbot (KarenBuildBot) is compiling Android builds of PPSSPP just fine, might just be an issue on Orphis's end?

[thedax]

It might be good to have a temporary backup buildbot until we figure out or hear what's wrong with Orphis'. It's been almost a week without a new Android build, I believe.

[bhavin192]

@xsacha
hrydgard used different method to exit for solving this #5418, but now it takes little more time (7 sec.) to exit on symbian also
Panic error code (errd) shows
Application closed:
PPSSPP
KERN-EXEC3

[xsacha]

@bhavin192 I'll see what I can do :)
Is the kern-exit a problem? Or just want it instant close?

[sum2012]

Thanks @angelXwind for the Android build

[bhavin192]

@xsacha
Kern-exec3 error is shown only when errrd is enabled, so it's not a problem. 7 sec is really more time for exit

[unknownbrackets]

Something with the lftp upload or something is timing out. I can see it happening (and after a successful build too), but I don't know why, and don't even know where it's uploading to.

-[Unknown]

[xsacha]

@bhavin192 I changed the exit. It was doing the shutdowns twice. It doesn't seem to have helped it quit faster. It's taking about 3 seconds to quit. It's doing it cleanly though (kills threads and then shuts down the ppsspp parts).

[bhavin192]

@xsacha
3 sec is correct
we can do one thing when user clicks exit minimise ppsspp and in background it will exit correctly ;)

[hrydgard]

@xsacha, if the "clean" exit is taking too much time, just do exit(0) like before I guess :) Exiting cleanly is nice but not really important as long as you do it after saving the config and not before...

@Orphis , any chance you could have a look at the Android build on the buildbot?

[hrydgard]

@solarmystic , yeah, that one will obviously be reverted soon if it can't be quick-fixed.

[hrydgard]

Thanks for all the reports guys, end of May isn't happening, mid-June perhaps.

[daniel229]

Is this #6093 going to be reverted or something? 7 games regression so far.

[GodEaterBoy]

Hrydgard god eater 2 1.4 new version is not work right when i play it in the game it force close. :(

[GodEaterBoy]

Pls fix it .

[sum2012]

@Orphis
Windows 32 bit buildbot broke

[thedax]

That's because 32-bit compilation is broken.

Error   43  error C2664: 'HRESULT IDirectInput8W::EnumDevices(DWORD,LPDIENUMDEVICESCALLBACKW,LPVOID,DWORD)' : cannot convert argument 2 from 'BOOL (__cdecl *)(LPCDIDEVICEINSTANCE,LPVOID)' to 'LPDIENUMDEVICESCALLBACKW'  \Windows\DinputDevice.cpp    124 1   PPSSPPWindows

[sum2012]

tag @Orphis
buildbot broke again

[thedax]

So what do we have left that's blocking us? Just curious, since it's been a while.

[unknownbrackets]

Well, a broken buildbot again which = insufficient testing of recent changes.

There's still at least two regressions from simulate block transfers.

-[Unknown]

[thedax]

Yeah, the buildbot goes without saying. Good to know about regressions, though.

[Joelgp83]

#6107 still exists in the current build on buildbot. Its definitely a regression since it used to work fine.

[ClubIce]

#6295 Fullscreen issue in Intel HD Graphics 4000

[sum2012]

@hrydgard when do v0.9.9 release ?

[GamerzHell9137]

@sum2012 When its ready and when most of the issues with games have been fixed.

[solarmystic]

The flickering issue in Breath of Fire 3, which was fixed by #6239 (v0.9.8-1045-g96787a9) has now returned again since v0.9.8-1220-g40f13d5 40f13d5 which means it is a regression.

#2759 (comment) #2759 (comment) #2759 (comment)

EDIT:- Fixed with #6383

[paricillt]

Please fix monster hunter portable 3rd on android I always get a black screen

[thedax]

That's not a regression. You need the PAUTH files (see the forums for more info) for it to run.

[Joelgp83]

Writing to confirm that #6107 is fixed on my device. Thanks @unknownbrackets! :)

[hrydgard]

I'm gonna restart the release process in new issue, this has grown way too old and big. Please re-test any regressions reported above and report them in the new one, #6422.

[sum2012]

@Yoshitsune
Do you mean this ?

[xsacha]

@sum2012 That's anisotropic filtering

[unknownbrackets]

It works fine at 1x. At 2x+, it looks like that. It's using s16 coords in throughmode, so potentially it might work adding a (res-1)res of a pixel to the coordinate or something. Although, I'm not actually sure why it doesn't just work. It is rotating UVs I think...

Anyway, I think it's always been that way, I don't think it's a regression?

-[Unknown]

[solarmystic]

It works just fine at 2x+ if you force Filtering to Nearest:-

[Yoshitsune]

What could be the problem then start?

[sum2012]

@Yoshitsune try set PSP Model 1000

[Yoshitsune]

@sum2012 strange, I did but still the problem.

[Yoshitsune]

There is nothing to worry about. hope in the latest version of ppsspp will be fixed

[echosen]

@hrydgard in the new update there is been some serious lag and frame rate issues with games that I have never had issues with before, such as FMA brotherhood, Gundam vs Gundam next plus, and fairy tale portable guild. I have tried several different setting to make the game run as smooth as before, also including certain experimental development options as I know of them from dealing with Open G.L scripts and O.S database. Ihave donated as well as I support your work, but I think when you are working on the next up date try to pick out the keys that are wrong and balance them with the ones that are right, because it seems with every update, something is tweaked which is a good thing but removing the entire issue makes it a problem.

In regards to Android PPSSPP