| |
hedning
Registered: Mar 2009
Posts: 4617 |
Release id #210333 : Babes
Discussion emerged. Posts below:
User Comment
Submitted by hedning [PM] on 17 October 2021
Looking forward to more tests btw. On real drives.
User Comment
Submitted by hedning [PM] on 17 October 2021
Ran it on a PAL 250469 REV A with a 1541-II without fastloader/cart. Cold boots between all runs. Tested it 20 times (even loaded different babes each time). Every load worked. I would say this release contains no loader bug (at least no bug that should matter), whatever you guys say, as it works, and should in emulation too. :)
And no fixed emulator version should be needed.
User Comment
Submitted by hedning [PM] on 17 October 2021
Copyfault: Making a modification should have its own entry imho. But I will try to run it from disk a couple of times too. If that works 100% the goal should be to emulate this behaviour in emulation too, even if it's a "bug" that makes it fail in Vice and 1541U2+. The goal for emulation is to mimic real hw. We should not make fixes for stuff to run on emulators. But I'll be back with a report.
User Comment
Submitted by Copyfault [PM] on 17 October 2021
Thanks for all your comments! To clear it up: on my real hw it refused to start most of the time! I did not yet try a real drive, so this might ease the pain.
Now the funny things after reading your proposals:
- it seems to always load with 1541-uII+ if I choose AR 6.0 as cart
- I did not ask to get it run for the actual contents but rather out of interest for such an edge case, since sometimes this sparks new ideas for exploitations
- even when I get this to load completely, it does always do a direct exit from the selection menu and the note is the only thing I get to see
The only real fix was the noloader-version by Ian Coog, since this can be loaded without any problems independant of the system setup (real thing/emu/cartridge setting) plus this one stays in the selection menu as one would expect. @IanCoog: thanks for fixing! Though I was sure this is some standard thing I have to admit that I did not know by heart the steps that need to be taken to get rid of the autostart stuff!
Would it be acceptable to add a dl link for this noloader-version to this entry? Or is this already "something new"?
User Comment
Submitted by Krill [PM] on 17 October 2021
Seems to be a race-condition, i.e., a bug in the code, not so much the emulator.
Set breakpoint in drive at $0403, then load with ,8,1.
When it breaks into the monitor,
vv vv vv
good >8:0000 0f 01 01 0f 01 00 00 00 1f 10 06 00 12 00 12 00
evil >8:0000 0f 01 80 0f 01 00 00 00 1f 10 06 00 12 00 12 00
Success depends on whether the block at T06/S00 has been loaded yet or not.
That it seems to always work on the real thing might be a meta-stability, and may depend a lot on various delays caused by physical disk access and disk layout.
That it reliably breaks in VICE (plain BASIC) might have to do with that.
I'd not be surprised if it randomly breaks on the real thing, too (with a somewhat lower probability than breaking in VICE).
User Comment
Submitted by Groepaz [PM] on 17 October 2021
It actually works in VICE (and chameleon too, likely also 1541) when loading it with AR5/6 fastload enabled. Very odd :)
(it's also not worth the trouble... pretty meh pictures :/)
User Comment
Submitted by Groepaz [PM] on 17 October 2021
How is it "faulty" though? it started up no problem 3 times in a row on the real thing :)
User Comment
Submitted by iAN CooG [PM] on 17 October 2021
Copyfault: that's the actual easier way to run it, by removing the faulty autoboot code, it's a simple diskeditor/hexeditor task.
https://www.dropbox.com/s/f0y6kc4oo30rm3n/Babes-noloader.D64?dl..
User Comment
Submitted by Groepaz [PM] on 17 October 2021
The most important question is - does it work on the real thing?
User Comment
Submitted by Copyfault [PM] on 17 October 2021
Stumbled over this after seeing the new comments lately. So I tried to run it on the real thing (with 1541U-II+) and on the latest vice emulator also - without sucess!
Either it just stops loading, or -in the rare cases it manages to finish loading and starts to decrunch- the selection screen is built up but left towards the note immediately, without any chance to select "Collection".
Any ideas what's going on here? Any chances to create a... let's say "fixed version"?
User Comment
Submitted by TheRyk [PM] on 12 October 2021
yeah some drive emulation killer, I also needed several tries
RRR!
User Comment
Submitted by hedning [PM] on 12 October 2021
Took me a couple of tries in Vice to get it running (run it without fastloader), but eventually it works. |
|
... 32 posts hidden. Click here to view all posts....
|
| |
Krill
Registered: Apr 2002
Posts: 2854 |
Quoting TheRykas already was stated, silly to do it just for this particular image, but should be possible in theory... Why? It's for science! To validate a theory! (And if proven right, the production would break.) :) |
| |
Copyfault
Registered: Dec 2001
Posts: 467 |
Quoting KrillQuoting TheRykas already was stated, silly to do it just for this particular image, but should be possible in theory... Why? It's for science! To validate a theory! (And if proven right, the production would break.) :) "For science". This! |
| |
Frantic
Registered: Mar 2003
Posts: 1629 |
We should all strive for wisdom, courage, temperance and justice, as Plato would have it. Pretty sure experiments with g64 images and alignment belong in there! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11148 |
meanwhile i have played around in VICE and made it generate the tracks with some offset similar to a real disk.... but no dice. there might be more to it :)
i also noticed that in the previous test the drive was running at just 287rpm, and it still worked fine - which makes it even less likely that stepping/seeking times are the problem |
| |
Krill
Registered: Apr 2002
Posts: 2854 |
Another thing to note is that Action Replay (which makes it work in VICE, unlike plain KERNAL) sets the disk controller timer period ($1c07) to $20 bycles before falling back to autostart KERNAL slowload, which then proceeds with that timer period without resetting it to the original value.
The original value is $3a bycles, so somewhat less frequent interrupts to check the disk until the M-E'd code sets it to $10 bycles (but then it's too late). |
| |
ChristopherJam
Registered: Aug 2004
Posts: 1381 |
Quote: Quote:And the reverse, perfectly aligned disk on realthing to break it. :)
but how? =)
I mean, you can get pretty damn close by setting a timer off the preceding track to tell you when to write the next one. It was on the infinite todo list until Krill offered to write it first.
(of course, a certain someone then got all sidetracked into writing a faster screen-off loader than the one I was working on instead, so... ;-) ;-) ) |
| |
chatGPZ
Registered: Dec 2001
Posts: 11148 |
https://sourceforge.net/p/vice-emu/code/40891/
doesnt really help with this collection though :) |
| |
Joe
Registered: Apr 2002
Posts: 224 |
Must be some very important collection making so many people discussing it and.. There is no graphician acredited for the original work?
Hope you get the intriguing disc-whatever-emulation/real-stuff working ;D
Laughing my arse off! |
| |
chatGPZ
Registered: Dec 2001
Posts: 11148 |
Quote:There is no graphician acredited for the original work?
gfx credits for the original work of converting some random tit pics? you must be kidding |
| |
Joe
Registered: Apr 2002
Posts: 224 |
SO YOU HAVE SEEN IT! Share it ;D
Haha, I didn't think it had any value, but it obviously did for different reasons. |
| Previous - 1 | 2 | 3 | 4 | 5 - Next |
