Dear IIe emulator authors [message #338193] |
Thu, 23 February 2017 09:21 |
zellyn
Messages: 173 Registered: April 2013
Karma: 0
|
Senior Member |
|
|
Hi folks,
You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
Right now the high RAM ("language card"-like functionality), auxmem, and softswitch tests are mostly done, and pretty useful. After that it goes into a set of color tests where it generates identical color patterns in two different ways, so you can compare them. (Some jumping of margins is expected)..
I intend to eventually add tests that the floating bus is reading the correct RAM locations on every cycle, and that the vertical blanking signal is cycle-accurate.
Any other bugs people point out to the IIe functionality in OpenEmulator are likely to become tests here too.
Hopefully this helps raise the bar for all IIe emulators out there. Although there are no serious problems, none of them currently pass all the language card functionality tests (except OpenEmulator, natch!).
Enjoy,
Zellyn
|
|
|
Re: Dear IIe emulator authors [message #338194 is a reply to message #338193] |
Thu, 23 February 2017 09:59 |
BLuRry
Messages: 489 Registered: October 2012
Karma: 0
|
Senior Member |
|
|
On Thursday, February 23, 2017 at 8:21:24 AM UTC-6, Zellyn wrote:
> Hi folks,
>
> You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
>
> Right now the high RAM ("language card"-like functionality), auxmem, and softswitch tests are mostly done, and pretty useful. After that it goes into a set of color tests where it generates identical color patterns in two different ways, so you can compare them. (Some jumping of margins is expected).
>
> I intend to eventually add tests that the floating bus is reading the correct RAM locations on every cycle, and that the vertical blanking signal is cycle-accurate.
>
> Any other bugs people point out to the IIe functionality in OpenEmulator are likely to become tests here too.
>
> Hopefully this helps raise the bar for all IIe emulators out there. Although there are no serious problems, none of them currently pass all the language card functionality tests (except OpenEmulator, natch!).
>
> Enjoy,
>
> Zellyn
That's fantastic. The difficulty I had with the 6502 test from Klaus is that I was never quite sure how to interpret it without using a debugger to see what it was doing at any given time. Is there a list of all the failure modes where it loops on itself -- or did you compile it with the debug feature that was later added?
Curious to hear more. Also, link to source please? ;)
-B
|
|
|
Re: Dear IIe emulator authors [message #338202 is a reply to message #338194] |
Thu, 23 February 2017 10:33 |
zellyn
Messages: 173 Registered: April 2013
Karma: 0
|
Senior Member |
|
|
On Thursday, February 23, 2017 at 9:59:19 AM UTC-5, BLuRry wrote:
>> You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
Yeah, his tests are much more low-level. This is intended to be tried once you have disk reading working, and want to get all the nit-picky details right.
The source is at the location I… um… apparently forgot to include in that message. Ouch.
https://github.com/zellyn/a2audit
Zellyn
|
|
|
Re: Dear IIe emulator authors [message #338203 is a reply to message #338193] |
Thu, 23 February 2017 10:34 |
zellyn
Messages: 173 Registered: April 2013
Karma: 0
|
Senior Member |
|
|
On Thursday, February 23, 2017 at 9:21:24 AM UTC-5, Zellyn wrote:
> You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
You also might be interested in me actually linking to it. https://github.com/zellyn/a2audit
Oops (thanks BLuRry!)
Zellyn
|
|
|
Re: Dear IIe emulator authors [message #338236 is a reply to message #338203] |
Thu, 23 February 2017 17:50 |
BLuRry
Messages: 489 Registered: October 2012
Karma: 0
|
Senior Member |
|
|
On Thursday, February 23, 2017 at 9:34:13 AM UTC-6, Zellyn wrote:
> On Thursday, February 23, 2017 at 9:21:24 AM UTC-5, Zellyn wrote:
>> You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
>
> You also might be interested in me actually linking to it. https://github.com/zellyn/a2audit
> Oops (thanks BLuRry!)
>
> Zellyn
Sweet! Jace has bugs apparently. Well that will make for an amusing bug hunt. E007 and E00A. Is this also running the other Klaus tests?
-B
|
|
|
|
|
Re: Dear IIe emulator authors [message #338256 is a reply to message #338193] |
Thu, 23 February 2017 21:16 |
|
Originally posted by: John Brooks
On Thursday, February 23, 2017 at 6:21:24 AM UTC-8, Zellyn wrote:
> Hi folks,
>
> You might be interested in this (very preliminary) attempt to create a comprehensive test suite for IIe emulators. The idea is to pick up where Klaus Dormann's 6502 tests leave off.
>
> Right now the high RAM ("language card"-like functionality), auxmem, and softswitch tests are mostly done, and pretty useful. After that it goes into a set of color tests where it generates identical color patterns in two different ways, so you can compare them. (Some jumping of margins is expected).
>
> I intend to eventually add tests that the floating bus is reading the correct RAM locations on every cycle, and that the vertical blanking signal is cycle-accurate.
>
> Any other bugs people point out to the IIe functionality in OpenEmulator are likely to become tests here too.
>
> Hopefully this helps raise the bar for all IIe emulators out there. Although there are no serious problems, none of them currently pass all the language card functionality tests (except OpenEmulator, natch!).
>
> Enjoy,
>
> Zellyn
Here are a few Disk II problem areas I've run into on IIGS emulators which probably also apply to other Apple II emulators:
1) Support for the 81 valid disk nibble values. For invalid nibbles, runs of 3+ zero bits should exhibit both random 1 flip as well as 'extra 1' bits being inserted into the 0 runs.
2) Support for 1/4 track head positioning, ideally tracking partial 1/4 tracks too and quantizing to 1/4 track boundaries.
3) Allow reading track data 1/4 track adjacent to where it was written: IE, writes cover 3x 1/4 tracks.
-JB
@JBrooksBSI
|
|
|
|
Re: Dear IIe emulator authors [message #338265 is a reply to message #338258] |
Thu, 23 February 2017 23:10 |
|
Originally posted by: John Brooks
On Thursday, February 23, 2017 at 7:44:46 PM UTC-8, Zellyn wrote:
> Disk operation is an explicit non-goal... for now. (But pull requests happily accepted!) I believe OE might do better than most though :-)
>
> Zellyn
Okay. So the goal of this test suite is comprehensive coverage of //e CPU, memory, and video functionality? Will it test game port, speaker & cassette I/O?
Do you know if a Disk II test suite exists elsewhere?
Thanks,
-JB
@JBrooksBSI
|
|
|
Re: Dear IIe emulator authors [message #338266 is a reply to message #338265] |
Thu, 23 February 2017 23:24 |
zellyn
Messages: 173 Registered: April 2013
Karma: 0
|
Senior Member |
|
|
I hadn't thought about speaker and other IO, actually. It started out with the goal of doing things that can be tested programmatically (which is why floating buffer read locations is on the to-do list), but then I started adding video stuff because I needed it :-) I think eventually everything will be in scope... I think I'd better start modularizing the code :-)
Zellyn
|
|
|
Re: Dear IIe emulator authors [message #338347 is a reply to message #338266] |
Fri, 24 February 2017 13:03 |
qkumba
Messages: 1584 Registered: March 2013
Karma: 0
|
Senior Member |
|
|
PR submitted.
Stylistically, may I suggest that error codes are extracted to a separate file? I had to search all files to know which was the next available error code for new LC tests. I didn't want to change all of the following numbers, since that will cause issues for existing user reports, so I just started at last+1.
|
|
|
Re: Dear IIe emulator authors [message #338348 is a reply to message #338347] |
Fri, 24 February 2017 13:17 |
zellyn
Messages: 173 Registered: April 2013
Karma: 0
|
Senior Member |
|
|
On Friday, February 24, 2017 at 1:03:34 PM UTC-5, Peter Ferrie wrote:
> PR submitted.
Thanks!
> Stylistically, may I suggest that error codes are extracted to a separate file? I had to search all files to know which was the next available error code for new LC tests. I didn't want to change all of the following numbers, since that will cause issues for existing user reports, so I just started at last+1.
They're extracted to here: https://github.com/zellyn/a2audit/blob/master/v0/index.md (which maps to zellyn.com/a2audit/v0) by this: https://github.com/zellyn/a2audit/blob/a2992b28d/audit/build #L17
I keep intending to write something that parses them better, allowing multi-line comments above the +perror line, and renumbers them automatically, but… :-)
|
|
|
|