Hello, not all is a dream!
I have a real Dragon 64 with an Alldream original cartridge, and I must say that it works flawlesly . . . it's simple great!
I can really debug and trace sentence by sentence, put breakpoints and the like, but only in the real world.
I dumped the cartridge to my PC (via Audacity, it's great too!) and then prepared a bat to work with XROAR.
I don't know why, but there works everything fine but the Trace command (it does nothing).
As I saw my Dragon tracing without problems, I thoght that maybe the ROM-BASIC where too different, so I dumped them too (both 32 and 64 mode)
Well, now I have a duplicate of my real world in XRoar but with Alldream it keeps on doing nothing when it is told to TRACE.
Has anyone tried to work with Alldream loading it as a cartridge at $C000?
and if so, has anyone succeded to debug a program step by step?
By the way, I have the original Alldream manual, in case anyone needs a copy, I will scan it as images and put it on this thread as soon as possible.
Any idea will be really very appreciated.
Pd. I use the DragonTools made by Rolf Michelsen, I don't know how I have managed to live without this great utility, thanks a lot, Rolf!!
Pere
ALLDREAM
Re: ALLDREAM
I doubt that I will be able to help but I do have one question; have you tried XRoar with the ALLDREAM ROM image in the archive, and if so, do you observe the same problem as the one you describe with your own image?
Re: ALLDREAM
Hello Alastair,
I verified (fc) the file in the archive against the copy of my cartridge and are identical.
I have then begun to install a lot of XRoar versions to try with.
I have used this bat to start each XROAR version:
=========================================
copy c:\XRoar\APL-ROMs\AllDream.rom dream.rom
copy c:\XRoar\ROMs\d64rom1V30.rom d64_1.rom
copy c:\XRoar\ROMs\d64rom2V30.rom d64_2.rom
xroar -machine dragon64 -extbas d64_1.rom -altbas d64_2.rom -load dream.rom -kbd-translate -tape-fast -lp-file c:\XRoar\output.txt
del *.rom
==========================================
These are the results of the tests:
XRoar v0.19 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.20 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.21 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.22 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.23 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.23a w32 - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.24 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.25 w32 - - - ERROR when LOADING a FILE from TAPE (all the tests have been made with the same wav file)
XRoar v0.25.1 w32 - - ERROR when LOADING a FILE from TAPE (all the tests have been made with the same wav file)
XRoar v0.26 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.27 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28 w64 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-09-22 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-09-23 w64 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-11-25 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.28.1 212-11-25 w64 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
So, it seems that XRoar could be preventing TRACE to work properly. I know that Trace in Alldream relays intensively on interrupts, but as there are some XRoar versions that work, not perfect but do something, for instance they let me add Breakpoints while the others don't, I'm thinking that maybe sixxie could have a look at that strange behaviour to find out, luckily, a way to work as the real Dragon does.
Thanks to anyone giving any idea to solve this!!
Pere
I verified (fc) the file in the archive against the copy of my cartridge and are identical.
I have then begun to install a lot of XRoar versions to try with.
I have used this bat to start each XROAR version:
=========================================
copy c:\XRoar\APL-ROMs\AllDream.rom dream.rom
copy c:\XRoar\ROMs\d64rom1V30.rom d64_1.rom
copy c:\XRoar\ROMs\d64rom2V30.rom d64_2.rom
xroar -machine dragon64 -extbas d64_1.rom -altbas d64_2.rom -load dream.rom -kbd-translate -tape-fast -lp-file c:\XRoar\output.txt
del *.rom
==========================================
These are the results of the tests:
XRoar v0.19 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.20 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.21 w32 - - - UNABLE TO LOAD THE ROM, so no way to test
XRoar v0.22 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.23 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.23a w32 - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.24 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.25 w32 - - - ERROR when LOADING a FILE from TAPE (all the tests have been made with the same wav file)
XRoar v0.25.1 w32 - - ERROR when LOADING a FILE from TAPE (all the tests have been made with the same wav file)
XRoar v0.26 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.27 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28 w64 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-09-22 w32 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-09-23 w64 - - - TRACE DOES NOTHING (just stops and shows the initial op-code but doesn't do any more)
XRoar v0.28.1 212-11-25 w32 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
XRoar v0.28.1 212-11-25 w64 - - - TRACE WORKS BUT randomly SKIPS over some SENTENCES (op-codes and parameters)
So, it seems that XRoar could be preventing TRACE to work properly. I know that Trace in Alldream relays intensively on interrupts, but as there are some XRoar versions that work, not perfect but do something, for instance they let me add Breakpoints while the others don't, I'm thinking that maybe sixxie could have a look at that strange behaviour to find out, luckily, a way to work as the real Dragon does.
Thanks to anyone giving any idea to solve this!!
Pere
Re: ALLDREAM
Hello,
I have given a try to DskDream which has the pro of being capable of working with drive units to the cost of loosing from $5E00 to $7FFF
The results have been the same that ALLDREAM from cartridge gave, on XROAR.
The versions that couldn't trace, now keep on being unable to.
The versions that traced but 'losing' randomly instructions, here do the same.
That reinforces the idea that there is something in XRoar that interferes with the Trace Option of these two packages.
Help, sixxie!!
Pere
I have given a try to DskDream which has the pro of being capable of working with drive units to the cost of loosing from $5E00 to $7FFF
The results have been the same that ALLDREAM from cartridge gave, on XROAR.
The versions that couldn't trace, now keep on being unable to.
The versions that traced but 'losing' randomly instructions, here do the same.
That reinforces the idea that there is something in XRoar that interferes with the Trace Option of these two packages.
Help, sixxie!!
Pere
Re: ALLDREAM
Have you tried the latest snapshot version of XROAR?
Re: ALLDREAM
Yes, I did.
From version 0.26 on trace doesn't work at all, but this snapshot recovers the old 'hald good' behavior and traces but not linearly, it jumps instructions at random
so it is better tat others, but . . .
PEre
From version 0.26 on trace doesn't work at all, but this snapshot recovers the old 'hald good' behavior and traces but not linearly, it jumps instructions at random
so it is better tat others, but . . .
PEre
Re: ALLDREAM
Are you running XROAR under Microsoft Windows?
Re: ALLDREAM
Yes,
I am using Windows 7 and it has always worked great.
This has to be a matter of interruptions. Alldream uses them a lot when one enters Dreambug.
Thanks for your interest
Pere
I am using Windows 7 and it has always worked great.
This has to be a matter of interruptions. Alldream uses them a lot when one enters Dreambug.
Thanks for your interest
Pere
Re: ALLDREAM
It would be interesting to see if the same thing happens with MESS.
Re: ALLDREAM
This is interesting... Can you give some steps to reproduce? I'm not sure what I'm looking for.
However it does indeed look interrupt related - Dream looks like it does tracing by waiting until an IRQ will trigger then returning to the next instruction to execute (using PULS CC,...,PC to unmask the IRQ).
In fact the way it waits until an IRQ *will* trigger looks very timing related (which would explain the differences between XRoar versions as I tweak timings). To the extent that it runs *differently* on the Tano (i.e. won't run properly without a patch). Interesting approach, though!
The idea seems to be to SYNC, then execute exactly as many instructions as needed such that the IRQ generated by field sync will register during the next instruction. So I can see that if things are off by even one cycle somewhere that some instructions (maybe the really short ones like NOP, etc.?) won't trace properly. Does this seem to tally with what you're seeing?
Edit: ok I have a test case, and that does seem to be basically what the problem is... now to somehow resolve that without breaking other timing issues!
However it does indeed look interrupt related - Dream looks like it does tracing by waiting until an IRQ will trigger then returning to the next instruction to execute (using PULS CC,...,PC to unmask the IRQ).
In fact the way it waits until an IRQ *will* trigger looks very timing related (which would explain the differences between XRoar versions as I tweak timings). To the extent that it runs *differently* on the Tano (i.e. won't run properly without a patch). Interesting approach, though!
The idea seems to be to SYNC, then execute exactly as many instructions as needed such that the IRQ generated by field sync will register during the next instruction. So I can see that if things are off by even one cycle somewhere that some instructions (maybe the really short ones like NOP, etc.?) won't trace properly. Does this seem to tally with what you're seeing?
Edit: ok I have a test case, and that does seem to be basically what the problem is... now to somehow resolve that without breaking other timing issues!