Oh Dear. I don't know which is worse, the fact you had a Skid Row track, or that you had to listen to it a thousand times :)
Darren Branagh Sent from My Android Phone. On 19 Dec 2016 12:37, "Marcel Kilgus" <ql-us...@mail.kilgus.net> wrote: > Wolfgang Lenerz wrote: > > Hi, > >> > >> Last night I tried it out on SMSQemulator and it worked much better > >> than anticipated from the get-go. > > Thanks ... I think. > > Hey, I don't usually expect things to work right on the first try, so > this was a pleasant surprise ;) > > > Anyway, if you set the SSSS frequency to 20 Khz in the config menu, (and > > if the version of java used can handle that frequancy) even the sound > > should play smoothly. > > Unfortunately 20kHz doesn't work for me at all. > > But I've spend days if not weeks getting the sound for QPC just right. > To this day I can't hear the song "18 and life" without thinking about > the coding sessions, because that was my test song and I've probably > heard it a thousand times back then, listening for every click and > scratch in the sound. > > In the end I let DirectSound do the frequency transformation, but I > also had versions where I did the sound output at 22Khz and upsampled > the 20kHz SSSS sound to it myself. > > > Yep, it loads the file in its "entirety" as per the documentation so it > > stands to reason that it resets the file pointer to the beginning, which > > is what the dv3 drivers do (file dv3_dv3_buff_asm, label dv3_sload). I > > remember that I didn't do that for SMSmulator in the beginning and got > > into a load of trouble. > > I actually didn't know this and the DOS device doesn't do it. But it's > not used as the main disc, so it can get away with more > incompatibilities than other drivers. > > >> "Yeah, but IOB.FMUL can handle at least 64KB!". Well, in many device > >> drivers the value in D1 is extended using "ext.l", so the trap will > >> fail to work as documented for anything above 32kb... > > Yes, that can be found in quite a number (all?) of i/o routines in > > Smsq/e. (Hastens to check that SMSQmulator is OK in that respect). I > > presume that it is easier to change the documentation in this regard, > > rather that trying to find that in every piece of code. > > Still it's amazing that such a elementary bug is in there for 30 years > or so and even I didn't know about it until two weeks ago. > > > If really necessary, it could be possible to create a new trap#3 (d0 > > could be $3f) for loading a full long word's worth as of the current > > file position. The changes for dv3 drivers would be minimal (just a few > > changes in dv3_dv3_io_asm). Any dv3 driver would automatically benefit > > from that. Normal drivers would come back with a not implemented error - > > software could check for that and use the normal trap if this didn't > > succeed. > > Sure. But on the other hand if nobody needed it for 30 years it's > probably not worth the trouble. ;) And I really like QMovie, I spent > many more hours on it than I could actually afford, but it's not worth > for that alone. And I've watched my Iron Maidens demo clip probably a > hundred times now during development, so I should probably move on to > other projects ;-) > > Cheers, Marcel > > _______________________________________________ > QL-Users Mailing List > _______________________________________________ QL-Users Mailing List
