Stella 4.6.5 released

[Keatah]

Good. Now I'm satisfied.

 

Maybe it has to do with leaving enough time to play sounds or flash the screen when other rollovers are hit. Or when you clear the three targets it makes that buwhoommp-buhwoommp-buhwoommp sound. It's something I wondered about off and on for years.

[+stephena]

The Stelladaptor/2600-daptor paddle issue is now fixed, and will be present in the next point release (later this week).

[+stephena]

I decided that since I have some work to do around the house for the next few days, I will do the new release now. So 4.6.7 is now available, with changes as follows:

 

* Fixed bug when using real paddles plugged into a Stelladaptor 2600-daptor device; the movement was very erratic.

* Fixed small logic error in 'MDM' bankswitching scheme.

 

As usual, comments, bug reports, etc can be done here, or by PM. Donations are welcome, and the download is available here.

[Thomas Jentzsch]

There is an annoying bug in the debugger: When Stella hits a break point, the scan line count (and related) are reset to 0.

[+stephena]

Is this on normal breakpoints (PC addresses), or more complicated conditional breakpoints, or both?

[Thomas Jentzsch]

I have only checked normal breakpoints.

[Thomas Jentzsch]

It seems that the CX22 is much less sensitive (50%) to movement than CX80 and Amiga mouse. Maybe you can consider this for Stella too.

[Avar]

Not 100% certain whether this is a Stella issue or a Blissbox issue, but Stella refuses to save input config for 2600 joysticks via BlissBox and it doesn't automatically work with them without config.

 

Edit: Fixed for joysticks. For some reason it finally worked after setting Compatibility to both admin -and- windows 7 (even though I'm in Windows 7) and simply selecting default.

 

However, issue remains for Keyboard Controller/Kid's Controller/Video Touch Keypad

Edited November 29, 2015 by Avar

[alex_79]

I found a bug in Stella emulation regarding the "playfield SCORE mode".

On real hardware, when the SCORE bit (D1 od CTRLPF) is set, the left half of the playfield gets color and priority of Player0, while the right half gets color and priority of Player1. In Stella, instead, the playfield gets the players' colors but keeps the same priority as the ball.

Here is a simple test rom to show the problem:
score_mode_test_NTSC.bin

score_mode_test_PAL.bin

In the test rom the left difficulty switch sets/resets the SCORE bit, while the right difficulty switch sets/resets the PFP bit (Playfield Priority, D2 of CTRLPF).
P0 = green vertical bars
P1 = blue vertical bars
BL = thin grey diagonal lines
PF = checkerboard pattern


Here are comparision pictures of the rom running in stella and on real hardware:
SCORE bit (left diff) set, PFP bit (right diff) unset:

Note how on real hardware P1 (in blue) in the left part of the screen is behind the PF but in front of the ball

SCORE bit (left diff) unset, PFP bit (right diff) unset:

both players in front of the playfield and ball. Correctly emulated in Stella.

PFP bit (right diff) set (SCORE bit/left diff doesn't matter in this case):

both players behind playfield and ball. Correctly emulated.

If we call S0 and S1 the left and right halves of the Playfield when the SCORE mode is enabled, then these are the correct priority assignements:

Normal priority (CTRLPF D2=0, D1=0)
Highest Priority: P0,M0
Second Highest: P1,M1
Third Highest: PF,BL
Lowest Priority: BK

PFP set (CTRLPF D2=1, D1=don't care)
Highest Priority: PF,BL
Second Highest: P0,M0
Third Highest: P1,M1
Lowest Priority: BK

SCORE mode (CTRLPF D2=0, D1=1)
Highest Priority: P0,M0,S0
Second Highest: P1,M1,S1
Third Highest: BL
Lowest Priority: BK

Edited December 31, 2015 by alex_79

[Thomas Jentzsch]

It seems that the CX22 is much less sensitive (50%) to movement than CX80 and Amiga mouse. Maybe you can consider this for Stella too.

This has been confirmed now.

[R.Cade]

Download links on the website are dead, so I assume there is an update forthcoming?

[Keatah]

They work fine.

[R.Cade]

Ah, you are correct. It seems sourceforge is on a lot of DNS blacklists as of late from past bad behavior.

[Keatah]

Probably fallout from the spyware they include with some downloads.

[Keatah]

Feature request: ability to change bankswitch on the fly, or at least without having to exit to the launcher - ok if a reset is done.

[dualcam]

Not sure that this is a really a Stella bug, but have ran into a paddle dead zone with 2600- and Stella-daptor's in linux (ubuntu). I'm pretty sure this is an analog joystick center dead zone, which I see when running jstest-gtk. I edit the calibration, setting center min/max values the same, and the dead zone is then gone from the jstest-gtk test screen, but still there in Stella. Searching found suggestions to add SDL_JOYSTICK_DEVICE=/dev/input/js0 to /etc/environment, but still getting the dead zone in Stella.

 

Anyone using paddles in Stella on Linux? I am surprised this has not come up before. Maybe a "feature" of SDL2?

 

Tom

http://2600-daptor.com

[dualcam]

Not sure that this is a really a Stella bug, but have ran into a paddle dead zone with 2600- and Stella-daptor's in linux (ubuntu). I'm pretty sure this is an analog joystick center dead zone, which I see when running jstest-gtk. I edit the calibration, setting center min/max values the same, and the dead zone is then gone from the jstest-gtk test screen, but still there in Stella. Searching found suggestions to add SDL_JOYSTICK_DEVICE=/dev/input/js0 to /etc/environment, but still getting the dead zone in Stella.

 

Anyone using paddles in Stella on Linux? I am surprised this has not come up before. Maybe a "feature" of SDL2?

 

Tom

http://2600-daptor.com

 

This looks to be a SDL2/linux problem. Was able to get the paddles working smoothly in Stella 3.9.3, last release prior to SDL2. Requires putting SDL_JOYSTICK_DEVICE=/dev/input/js0 to /etc/environment, and using jstest-gtk to edit the calibration and set all center min/max values to 2047.

 

Going to Stella 4.0 and SDL2, and the paddle dead zone is back. Appears linux applies a calibration to anything that is an analog joystick, and the default calibration includes an axis center dead zone. Don't know of a way to change the calibration on the /dev/input/eventX that SDL2 uses - jstest-gtk seems to only work on /dev/input/js0. And SDL2 seems to be ignoring the SDL_JOYSTICK_DEVICE=/dev/input/js0.

 

Tom

http://2600-daptor.com

[+stephena]

I found a bug in Stella emulation regarding the "playfield SCORE mode".

 

On real hardware, when the SCORE bit (D1 od CTRLPF) is set, the left half of the playfield gets color and priority of Player0, while the right half gets color and priority of Player1. In Stella, instead, the playfield gets the players' colors but keeps the same priority as the ball.

 

Confirmed, and currently working on fixing this for the next release.

[+stephena]

Feature request: ability to change bankswitch on the fly, or at least without having to exit to the launcher - ok if a reset is done.

 

What is the use case for this? I suppose it's easy enough to handle; just cycle through the bankswitch schemes and do a reload, all tied to a key-combo. I just don't see the point of it.

[Keatah]

Key combo.. I'll try that. I had some roms not working till I tried about 10 schemes before getting it right.

[+stephena]

To be clear, there's no key-combo right now to do it. I was just saying that if/when I implement the feature, it would be tied to a key-combo. I'll look into it for the next release.

[+stephena]

I found a bug in Stella emulation regarding the "playfield SCORE mode".

 

On real hardware, when the SCORE bit (D1 od CTRLPF) is set, the left half of the playfield gets color and priority of Player0, while the right half gets color and priority of Player1. In Stella, instead, the playfield gets the players' colors but keeps the same priority as the ball.

 

This is now fixed, and will be present in the next release.

[+stephena]

It seems that the CX22 is much less sensitive (50%) to movement than CX80 and Amiga mouse. Maybe you can consider this for Stella too.

 

Do you have an idea of how to fix this? I ask because the relevant code is from z26, which you probably worked on at some point.

[Thomas Jentzsch]

Do you have an idea of how to fix this? I ask because the relevant code is from z26, which you probably worked on at some point.

Unfortunately I only tested what Eckhard had developed.

 

But from looking at the code:

void TrackBall::update()
  ...
  myTrakBallCountH = abs(myVCounter >> 1);
  myTrakBallCountV = abs(myHCounter >> 1);

probably a right shift by 2 (instead of 1) in case of a CX22 would do.

But those shifts (even the existing ones) create a problem: Very slow mouse movements are ignored completely. But only in case of positive values (1 >> 1 becomes 0; -1 >> 1 is still -1 (is it?)), so that the cursor would have a preference into left or right and up or down. Not sure if it becomes obvious now. If that's the case, the counters would have to be adjusted before shifting in case of positive values.

[+stephena]

 

Unfortunately I only tested what Eckhard had developed.

 

I'm going to leave this alone for now, as it needs more research and I'm too close to the next release. But I'd appreciate you having a look and fixing it, if possible