28-Mar-2024 23:28 GMT.
UNDER CONSTRUCTION
Anonymous, there are 155 items in your selection [1 - 50] [51 - 100] [101 - 150] [151 - 155]
[News] Interview with Thomas Frieden: OS4 questions and answersANN.lu
Posted on 10-Sep-2002 12:19 GMT by Teemu I. Yliselä155 comments
View flat
View list
"A few weeks ago, the amiga-news.de community collected a bunch of questions related to OS4 and forwarded them to Thomas Frieden, in the hope that he may answer some of them. Guess what? Not only did he answer all of them, he even answered them in English, for you poor souls outside of Germany ;-)" Read the interview over at amiga-news.de here.
Interview with Thomas Frieden: OS4 questions and answers : Comment 1 of 155ANN.lu
Posted by brotheris on 10-Sep-2002 11:46 GMT
"Will OS4 automatically recognise if a floppy disk gets inserted?
As this is not yet implemented, I can't really say. However, floppy support is becoming more and more unimportant right now, so I don't think too much energy will be wasted on such a dead medium. "
This concernes other removable media too.
Interview with Thomas Frieden: OS4 questions and answers : Comment 2 of 155ANN.lu
Posted by EmGee on 10-Sep-2002 11:57 GMT
In reply to Comment 1 (brotheris):
Many other removable media use the IDE/SCSI interface, not the floppy interface. So it's a mather of drivers i guess.
Interview with Thomas Frieden: OS4 questions and answers : Comment 3 of 155ANN.lu
Posted by johnny_boy on 10-Sep-2002 12:17 GMT
"Will the annoying "Disk not validated"-Problem be cured with the new implementation of FFS?
No."
This is BAD news. Invalid partitions are the reason I banished FFS from my hard drives years ago. I hope somebody comes up with an alternative PPC filesystem quickly.
Interview with Thomas Frieden: OS4 questions and answers : Comment 4 of 155ANN.lu
Posted by Peter Gordon on 10-Sep-2002 12:34 GMT
In reply to Comment 3 (johnny_boy):
Filesystems themselves are not "PPC", the implementation of the filesystem driver is either "PPC" or something else. Therefore, if SFS is written in 100% C it shouldn't be too much effort to recompile it for PPC.
The question is, do you want to run a filesystem that has just been ported to PPC? I'd wait until it had been beta tested by people who care a little less about their data, first ;)
<speculation type="wild"> From what I understand of the way OS4 handles 68k code, there shouldn't be any reason why you can't use 68k filesystem handlers to read/write your partitions, and while that would incur an extra overhead, it should still be much faster than it was on your 680x0. </speculation>
Interview with Thomas Frieden: OS4 questions and answers : Comment 5 of 155ANN.lu
Posted by Alkis Tsapanidis on 10-Sep-2002 12:42 GMT
In reply to Comment 4 (Peter Gordon):
Well, SFS was ported to MorphOS in very little time, and it's STABLE, I can't
think of any reason for this not to be possible with OS4.
Interview with Thomas Frieden: OS4 questions and answers : Comment 6 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 10-Sep-2002 13:00 GMT
In reply to Comment 5 (Alkis Tsapanidis):
In actual fact Jörg (who currently maintains SFS) is an OS 4 developer and he has already agreed to port SFS to OS 4.
Interview with Thomas Frieden: OS4 questions and answers : Comment 7 of 155ANN.lu
Posted by Jim Forbes-Ritte (AGAfaster) on 10-Sep-2002 13:30 GMT
In reply to Comment 4 (Peter Gordon):
-Filesystems implemented in m68k.
Hmmm. Interesting - on the face of it you are definitely right certainly given overheads, etc the speed should be pretty good.
however (Uh oh - cold water on the fireworks !) isnt the 68k emulated stuff being run in a seperate MMU space (the PPC MMU, not an emulated 68851, cos they aint doing that) ?
I'm not sure if the Filesystem running under Petunia would be available for the PPC side of things - it could work given that new functionality is available for 68k Apps, so it might work in vice versa.
but then again, I could be talking a load of old ar5e.
Interview with Thomas Frieden: OS4 questions and answers : Comment 8 of 155ANN.lu
Posted by TBone on 10-Sep-2002 13:41 GMT
I'd like to see the new OpenBeos implementation of the BeFS available, it's 64bit, journaled, database-like, multithreaded, bells, whisles, balloons, ect.
http://www.bug-br.org.br/openbfs/
Interview with Thomas Frieden: OS4 questions and answers : Comment 9 of 155ANN.lu
Posted by Don Cox on 10-Sep-2002 13:44 GMT
In reply to Comment 3 (johnny_boy):
"This is BAD news. Invalid partitions are the reason I banished FFS from my hard drives years ago. I hope somebody
comes up with an alternative PPC filesystem quickly."
In that case you may not realise how much the problem has been reduced
in OS 3.9.
It can still happen under 3.9 that the bitmap isn't correctly written
if the computer crashes during a write, but the validation process now
proceeds much more smoothly than in the past, and in my experience you
now only get an unvalidated drive if there is a total crash during a
write.
Even if a drive is not validated, you can still read from it, so any
recent files that have not been backed up can be rescued.
Interview with Thomas Frieden: OS4 questions and answers : Comment 10 of 155ANN.lu
Posted by Jim Forbes-Ritte (AGAfaster) on 10-Sep-2002 13:51 GMT
In reply to Comment 9 (Don Cox):
... not to mention little utilities like Disksafe.
This waits for a reset (usually happens after a crash) and writes the pending Bitmap (if any) to the disk if a disk write is interrupted.
I use this, and it has decimated Disk invalidation errors (these were usually caused by Ibrowse failing while writing to the cache) on my OS3.9 A1200. (OS3.9 sorted a lot of these too)
I Imagine this wont be long appearing on OS4.0 !
Interview with Thomas Frieden: OS4 questions and answers : Comment 11 of 155ANN.lu
Posted by Ole-Egil on 10-Sep-2002 13:59 GMT
In reply to Comment 10 (Jim Forbes-Ritte (AGAfaster)):
On the other hand, some of us haven't had a problem with this in a LONG time. I mean, the only times I've managed to crash my amiga during a write the last year or so has been while downloading some HUGE file over a slow network while trying to learn coding. In hindsight it could be said this is not a generally good idea in an environment like AmigaOS 3.x ;-)
Interview with Thomas Frieden: OS4 questions and answers : Comment 12 of 155ANN.lu
Posted by Olaf Barthel on 10-Sep-2002 14:19 GMT
In reply to Comment 3 (johnny_boy):
> "Will the annoying "Disk not validated"-Problem be cured with the new implementation of FFS?
>
> No."
>
> This is BAD news. Invalid partitions are the reason I banished FFS from my hard drives years
> ago. I hope somebody comes up with an alternative PPC filesystem quickly.
You can't eat your cake and have it, too. The FFS data structure design is very verbose,
by its very nature the consistency must be assured at all times. This is what the validation
process is for. There will always be a certain chance that if your system goes down at
the wrong moment, the subsequent restart of the file system will perform a consistency
check first. I'm happy to say that the new FFS implementation is more paranoid than the
original version and will both take greater care in finding/reporting problems and avoiding
them in the first place. The catch is that with the FFS data structures, in order to be
safer, you will have to be slower, too :(
Side note: the point in rewriting FFS to run on the new platform is not so much in delivering
top notch file system performance (FFS isn't cut out to be great in this respect) as to
allow for a certain degree of backwards compatibility and for data interchange. We definitely
need to do better in the performance department, but that's a different story altogether.
Interview with Thomas Frieden: OS4 questions and answers : Comment 13 of 155ANN.lu
Posted by Olaf Barthel on 10-Sep-2002 14:23 GMT
In reply to Comment 9 (Don Cox):
> [..]
> It can still happen under 3.9 that the bitmap isn't correctly written
> if the computer crashes during a write, but the validation process now
> proceeds much more smoothly than in the past, and in my experience you
> now only get an unvalidated drive if there is a total crash during a
> write.
>
> Even if a drive is not validated, you can still read from it, so any
> recent files that have not been backed up can be rescued.
That's correct. A lot of effort went into making the FFS reimplementation
multithreaded so as to allow (among other thing) to allow for file system
operations to be handled while the validation process is running.
Interview with Thomas Frieden: OS4 questions and answers : Comment 14 of 155ANN.lu
Posted by Ian on 10-Sep-2002 14:44 GMT
Some revealing answers which bring up some more questions in my mind:
The scheduling is the standard priority round-robin thing, which is fine except the load-dependent behaviour offered by Executive makes the system much better and this would have been an ideal opportunity to add it.
It is still unclear whether existing 3rd-party 68K-based filesystems (like PFS) will operate under OS4.
Floppy disk low priority? Understandable, but WinCE has no floppy support and that makes it a pain in the ass for some. Some limited support would really be desirable.
What about Mass Storage System USB drivers, for access to digital cameras?
Screen dragging may be hard/impossible, but will we retain multiple (non-draggable) screens?
What system-level redirects will be implemented for backwards compatibility? For example, system-friendly sound programs may open audio.device and stick to proper device commands. Will these be redirected to AHI, or will the program need to be redone? The same applies to other system-level access schemes to "old" hardware like joysticks.
I agree UAE is the better system to get old games/apps running rather than some convoluted hardware access hack (although such a hack would be nice...) , but will there be a UAE compile for OS4? Will there be any integration?
Not to diss what has been done! Other than those unclear points, the whole project sounds very promising. :-)
Interview with Thomas Frieden: OS4 questions and answers : Comment 15 of 155ANN.lu
Posted by johnny_boy on 10-Sep-2002 15:10 GMT
In reply to Comment 4 (Peter Gordon):
> Filesystems themselves are not "PPC", the implementation of the filesystem
> driver is either "PPC" or something else.
Yes, I didn't go in to that much detail in my quick whinge. :)
> The question is, do you want to run a filesystem that has just been ported to
> PPC?
If it was a port of SFS by it's original developer then I'd be prepared to trust it. SFS has proven to be reliable on my system when I've used it, although I found PFS to be quicker. You shouldn't trust any filesystem 100%, that's why we all make backups, right?
[68k filesystem handler under OS4 68k emulation]
If it works then PFS, for example, should certainly perform no worse than it does on a '060, given a fast enough PPC. But whether this will work at all is something that we can only speculate on until somebody can try it out for real.
Interview with Thomas Frieden: OS4 questions and answers : Comment 16 of 155ANN.lu
Posted by Bill T. on 10-Sep-2002 15:22 GMT
In reply to Comment 14 (Ian):
>Floppy disk low priority? Understandable, but WinCE has no floppy support and that makes it a pain in the ass for some.
>Some limited support would really be desirable.
Well, since they say the standard AmigaOne floppy controller can't read Amiga format disks
anyway, why worry about it so much until a Catweasel for PCI comes out? Though someone else
did make the point that it could also affect other removable media, such as ZIP, JAZ,
CDROM, etc. Hopefully the lack of interest in floppy at the moment won't hinder the usability
of these other removable formats.
Interview with Thomas Frieden: OS4 questions and answers : Comment 17 of 155ANN.lu
Posted by johnny_boy on 10-Sep-2002 15:23 GMT
In reply to Comment 9 (Don Cox):
[Invalid FFS partitions]
> In that case you may not realise how much the problem has been reduced
> in OS 3.9.
I was using the OS3.5 version when I decided that I'd had enough of the problem. I lost data as sometimes the validator program would crash before it was completed, the only fix being a format. That was when I decided that FFS was nolonger for this world. I know that FFS can be setup to perform well when compared to the likes of SFS and PFS, given large block sizes and a big buffer, but PFS gives that performance without the disk space/memory trade off.
I use OS3.9 FFS on a Games partition as some of my creaking old games require it, but that's not really demanding enough to be able to prove that the problems have been improved.
Perhaps it has, but frankly the phrase "once bitten, twice shy" springs to mind.
Interview with Thomas Frieden: OS4 questions and answers : Comment 18 of 155ANN.lu
Posted by dammy on 10-Sep-2002 15:28 GMT
In reply to Comment 9 (Don Cox):
How does that stack up against ext3?
Dammy
Interview with Thomas Frieden: OS4 questions and answers : Comment 19 of 155ANN.lu
Posted by Bill "tekmage" Borsari on 10-Sep-2002 15:35 GMT
In reply to Comment 15 (johnny_boy):
I trust vxfs 100% :)
I personally run a mix of FFS and SFS on my machines though I much prefer SFS. Funny that last night I had a crash of V during a write to a SFS partition. SFScheck reported it as damaged and sfssalv crashed attemping to clean it up. I'll see tonight if I can fix it.
Interview with Thomas Frieden: OS4 questions and answers : Comment 20 of 155ANN.lu
Posted by johnny_boy on 10-Sep-2002 15:47 GMT
In reply to Comment 12 (Olaf Barthel):
> You can't eat your cake and have it, too.
You can with PFS! ;)
> The catch is that with the FFS data structures, in order to be
> safer, you will have to be slower, too :(
> the point in rewriting FFS to run on the new platform is not so much in
> delivering top notch file system performance as to allow for a certain degree
> of backwards compatibility
Why does backwards compatibility in OS4 have to be achieved through developing FFS? It doesn't seem worth developing if no speed gains can be made.
If the OS3.9 FFS handler could only work by being ported to PPC, then why not do the minimum to get it working just for compatibility? Then implement a whole new reliable AND fast filesystem for OS4.
If a 68k handler will work through the OS4 emulation then all the better! Can you shed any light on this Olaf?
Interview with Thomas Frieden: OS4 questions and answers : Comment 21 of 155ANN.lu
Posted by Don Cox on 10-Sep-2002 15:56 GMT
In reply to Comment 17 (johnny_boy):
"I was using the OS3.5 version when I decided that I'd had enough of the problem. I lost data as sometimes the validator
program would crash before it was completed, the only fix being a
format. "
That crash is one of the things that has been fixed. It was indeed a
bad bug.
Interview with Thomas Frieden: OS4 questions and answers : Comment 22 of 155ANN.lu
Posted by Don Cox on 10-Sep-2002 16:07 GMT
In reply to Comment 20 (johnny_boy):
"If the OS3.9 FFS handler could only work by being ported to PPC, then why not do the minimum to get it working just for
compatibility? Then implement a whole new reliable AND fast filesystem
for OS4. "
How stable do you think an all new file system sitting on top of a
massively changed OS would be? It's taken about 18 years to get FFS to
the level of stability we have now. PFS and SFS have also had many
years of development and AFAIK are not 100% safe yet.
By all means let's have a new file system with all the latest ideas,
but I would only want to use it on my browser cache partition for the
first five years, not have it as the default.
BTW after 9 months of heavy use, I have not yet had an unvalidated
drive on my Amithlon machine. The A4000 OTOH is much less reliable and
has to validate the YAM partition regularly.
Interview with Thomas Frieden: OS4 questions and answers : Comment 23 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 10-Sep-2002 17:27 GMT
In reply to Comment 20 (johnny_boy):
The idea behind FFS2 is to provide a solid backwards compatibility option so people can attach existing harddisks and read from them.
Interview with Thomas Frieden: OS4 questions and answers : Comment 24 of 155ANN.lu
Posted by Olaf Barthel on 10-Sep-2002 17:59 GMT
In reply to Comment 20 (johnny_boy):
>> You can't eat your cake and have it, too.
>
> You can with PFS! ;)
Whatever rocks your boat...
> > The catch is that with the FFS data structures, in order to be
> > safer, you will have to be slower, too :(
>
> > the point in rewriting FFS to run on the new platform is not so much in
> > delivering top notch file system performance as to allow for a certain degree
> > of backwards compatibility
>
> Why does backwards compatibility in OS4 have to be achieved through developing
> FFS?
Migrating an existing installation becomes easier if you can plug in a disk and
just copy the data over. Now you'd say you could do that with the emulated 68k
FFS. That may be true, question is how fast it would be compared to the FFS
reimplementation and whether the media support would be as robust as the one
the reimplemented FFS has to offer.
> It doesn't seem worth developing if no speed gains can be made.
They can be made and they were made. It's just that the trade-off between
speed and reliability is particularly great with the FFS. It was not designed
to be robust with regard how data corruption is handled. SFS and in general
all journaling file systems are designed to handle this eventuality. With
FFS all you can expect is that you will be able recover data. This, however,
becomes less likely the more you push it towards being fast at writing.
> If the OS3.9 FFS handler could only work by being ported to PPC, then why not
> do the minimum to get it working just for compatibility? Then implement a whole
> new reliable AND fast filesystem for OS4.
Thank you for reminding me of the task I didn't feel qualified to handle
until I had done the FFS reimplementation ;) Among many other things, the
implementation effort also served as the means for me to learn how an
Amiga file system should be implemented. I learned a lot in the process,
not all of which I really wanted to know.
> If a 68k handler will work through the OS4 emulation then all the
> better! Can you shed any light on this Olaf?
Unless a file system expects certain arcane and undocumented BCPL kernel
functionality to be present in AmigaDOS (which it shouldn't), it should
continue to work on OS4.
Interview with Thomas Frieden: OS4 questions and answers : Comment 25 of 155ANN.lu
Posted by Mike Veroukis on 10-Sep-2002 18:52 GMT
In reply to Comment 9 (Don Cox):
>It can still happen under 3.9 that the bitmap isn't correctly written
>if the computer crashes during a write, but the validation process now
>proceeds much more smoothly than in the past, and in my experience you
>now only get an unvalidated drive if there is a total crash during a
>write.
True, but more importantly, OS4 has memory protection and if it works fairly well, a mis-behaved app that would normally crash an OS3.9 system would instead be halted by the OS. Thus, the filesystem would not be subjected to so many mid-write errors. Also, I wonder if OS 4 has a formal shut down procedure. If so, and if Amiga users get used to the idea, no one will simply power off the computer while a write is in progress. Even better would be if the OS could trap the power switch (assuming ATX power supply/MB) and it could hold off the shut-off until all I/O is complete.
- Mike
Interview with Thomas Frieden: OS4 questions and answers : Comment 26 of 155ANN.lu
Posted by Hooligan/DCS on 10-Sep-2002 18:57 GMT
I can fairly say that I have been using SFS since the first beta versions and I have had NOT EVEN ONE problem with it. And I am talking years here. Great piece of work.
Interview with Thomas Frieden: OS4 questions and answers : Comment 27 of 155ANN.lu
Posted by Jon on 10-Sep-2002 18:57 GMT
In reply to Comment 19 (Bill "tekmage" Borsari):
Uh-oh..try to localize damaged files and delete them (the dir) if you can. There is no known way (to me) to repair SFS partitions.
SFSCheck seems to find problems but can't do anything for them :(
Interview with Thomas Frieden: OS4 questions and answers : Comment 28 of 155ANN.lu
Posted by Jon on 10-Sep-2002 18:58 GMT
In reply to Comment 27 (Jon):
Also having a separate cache partition for crashy browsers is an excellent idea..
Interview with Thomas Frieden: OS4 questions and answers : Comment 29 of 155ANN.lu
Posted by Jon on 10-Sep-2002 19:02 GMT
In reply to Comment 25 (Mike Veroukis):
A Frieden told once, that there is no need to "shutdown".. also for the guy who asked for screens: where would they go? They already work with RTG systems..Only the dragging is a problem.
But is it (dragging the screens) useful anymore? Cool maybe..Virtual screens might be more powerful? Matter of taste.
Interview with Thomas Frieden: OS4 questions and answers : Comment 30 of 155ANN.lu
Posted by brotheris on 10-Sep-2002 19:32 GMT
In reply to Comment 25 (Mike Veroukis):
Protecting free memory won't do a lot of good to system stability. Current AOS design doesn't allow this.
Interview with Thomas Frieden: OS4 questions and answers : Comment 31 of 155ANN.lu
Posted by tinman on 10-Sep-2002 19:42 GMT
In reply to Comment 30 (brotheris):
Why not - you won't corrupt the free memory lists which is one surefire way to crash the current AOS.
Interview with Thomas Frieden: OS4 questions and answers : Comment 32 of 155ANN.lu
Posted by johnny_boy on 10-Sep-2002 20:09 GMT
In reply to Comment 22 (Don Cox):
> How stable do you think an all new file system sitting on top of a
> massively changed OS would be? It's taken about 18 years to get FFS to
> the level of stability we have now.
It depends on the quality of the design, coding and testing. You can't just say flatly that a new filesystem would be unreliable because the OS is being rewritten. That would mean you are saying that OS4 is going to be riddled with bugs and crash happy. I think that is doing a disservice to the talents of the people working on OS4. It's not likely that Hyperion would be happy to release OS4 if it was that bad.
In those 18 years OFS/FFS has had hardly any development. It only became a priority for development when hard drives became a standard component of a computer. Then there were 6 years of no development before OS3.5, so it's not like it takes nearly two decades to produce a reliable filesystem.
From what Olaf has said it would appear that FFS is being developed for a practice run just as much as for compatibility, which given the limited resources available for OS4 development is something we have to accept.
> PFS and SFS have also had many years of development and AFAIK are not 100%
> safe yet.
I don't believe that any filesystem is 100% safe, and no hard drive is. So I always make backups whether I encounter problems with a particular filesystem or not.
Interview with Thomas Frieden: OS4 questions and answers : Comment 33 of 155ANN.lu
Posted by Arild Kvam on 10-Sep-2002 20:20 GMT
In reply to Comment 16 (Bill T.):
Just use crossdos to copy files between Classic and AmigaOne then.
I don't see any reason why ZIP/JAZ/CD-ROM's wouldn't work on the AmigaOne just
because original amiga floppies doesn't.
Interview with Thomas Frieden: OS4 questions and answers : Comment 34 of 155ANN.lu
Posted by JoannaK on 10-Sep-2002 21:40 GMT
In reply to Comment 31 (tinman):
I take this as a good start. Protected system on Amiga has been on hard thinking and long discussions among users/developers and it's not easy thing to do. Especially if older programs are to be kept compatible with system. But even limited system is useful when debugging. (Like Enforcer.. :)
I really hope OS4 team have tested this clever MMU-based system (as explained by B.Hermans on thread " AmigaOS4 status update") using some test software before basing OS4 development on it. It would be quite catastrophic to discover that it does not work on real PPC-code. Afterall.. it was only couple months ago (june) Alan@Eyetech claimed on Amiga-one mailing list that Aone HW is quite capable of reading classix 880k disks. Now, on this latest interview Mr Frienden says it's not possible.
So, please tell me this method is tested on real PPC:s and it works without addign too much overhead to system use?
Interview with Thomas Frieden: OS4 questions and answers : Comment 35 of 155ANN.lu
Posted by strobe on 11-Sep-2002 03:18 GMT
Not much meat here. The primary question NOT answered is: WHEN
Interview with Thomas Frieden: OS4 questions and answers : Comment 36 of 155ANN.lu
Posted by Ole-Egil on 11-Sep-2002 03:59 GMT
In reply to Comment 34 (JoannaK):
To be QUITE frank, some people out there claim to use PC floppy controllers to _read in_ a floppy that has been formatted on an amiga. But whether this actually counts as "yes" or "no" in the discussion of "will I be able to use amiga floppies" is open to a LOT of debate. I think both answers are right, it just depends on how you define "use" ;-)
http://www.oldskool.org/disk2fdi/ for anyone who hasn't seen it yet. DOS only for now, and requires 2 floppies. The theory is stupid, but works ;-)
Interview with Thomas Frieden: OS4 questions and answers : Comment 37 of 155ANN.lu
Posted by Olaf Barthel on 11-Sep-2002 04:56 GMT
In reply to Comment 30 (brotheris):
> Protecting free memory won't do a lot of good to system stability. Current
> AOS design doesn't allow this.
I won't do a lot of good, but it could help a little bit; and that little bit
could make the difference. It's hard enough to detect random memory trashing in
memory that has been allocated by applications. But most of the time you'll have
plenty of unused memory around which could be trashed just as easily. And it
should be easier to monitor the state of the free memory than the memory
allocated by running applications for themselves and for others to share. As for
current AmigaOS design not permitting free memory to be protected, that's
certainly the case. Or at least, it's hard to do considering that the granularity
of memory chunks is not the same as the granularity of the MMU pages. How could
you solve this? By wasting memory (current AmigaOS) or by changing the memory
allocator in the first place.
Interview with Thomas Frieden: OS4 questions and answers : Comment 38 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 11-Sep-2002 05:33 GMT
In reply to Comment 37 (Olaf Barthel):
>By wasting memory (current AmigaOS) or by changing the memory
>allocator in the first place.
Which exactly what we did ofcourse. Exec SG now allows for different memory allocation schemes and the currently implemented allocation scheme in Exec SG is much more sophisticated than what we have now.
Interview with Thomas Frieden: OS4 questions and answers : Comment 39 of 155ANN.lu
Posted by Don Cox on 11-Sep-2002 05:33 GMT
In reply to Comment 24 (Olaf Barthel):
"Migrating an existing installation becomes easier if you can plug in a disk and
just copy the data over. Now you'd say you could do that with the emulated 68k
FFS. That may be true, question is how fast it would be compared to
the FFS
"
Judging from Amithlon, extremely fast. The modern IDE hardware is much
faster than the A4000 IDE.
Admittedly the software in an emulation running on a G3 will not be as
fast as in one running on an Athlon 1700, but the hardware is probably
the dominating factor here.
Interview with Thomas Frieden: OS4 questions and answers : Comment 40 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 11-Sep-2002 05:36 GMT
In reply to Comment 39 (Don Cox):
Small detail Don: FFS2 will run PPC native.
I'm pretty confident that PPC native code executed at 600 Mhz will be on par with emulated 68K code on a x86 box.
Still, I would agree that the determining factor here is the hardware and since the AmigaOne has a UDMA 100 controller, I don't see much problems.
Interview with Thomas Frieden: OS4 questions and answers : Comment 41 of 155ANN.lu
Posted by Don Cox on 11-Sep-2002 05:38 GMT
In reply to Comment 25 (Mike Veroukis):
"Also, I wonder if OS 4 has a formal shut down procedure. If so, and if Amiga users get used to the
idea, no one will simply power off the computer while a write is in
progress."
I hope it does not have a formal shutdown procedure. That is one of
the major annoyances in other OSes.
Now that disk writes are so fast, there is much less chance of a file
being partly written than in the days of 40Meg drives.
In any case, the problem is crashes and power failures rather than
switching off the computer at the end of a session.
Interview with Thomas Frieden: OS4 questions and answers : Comment 42 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 11-Sep-2002 05:40 GMT
In reply to Comment 34 (JoannaK):
The overhead with the MMU based exception scheme is quite low, even on a Cyberstorm.
PPC exception are far less CPU intensive than 68K exceptions.
We obviously considered all facts before making this decision.
The main reason behind it is according to Thomas that it obviates the need to insert emulation traps into source-code and it allows to keep the ABI/API "clean".
68K emulation will gradually become less important and we don't want legacy support to affect the design of Exec SG.
Interview with Thomas Frieden: OS4 questions and answers : Comment 43 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 11-Sep-2002 05:41 GMT
In reply to Comment 41 (Don Cox):
No, OS 4 does not have a formal shut-down procedure.
VM swapping will be done using a partition rather than a swap-file.
This obviates the need for a formal shut-down although obviously it cannot be advisable to turn off your computer whilst it is writing or reading from HD.
Interview with Thomas Frieden: OS4 questions and answers : Comment 44 of 155ANN.lu
Posted by Don Cox on 11-Sep-2002 05:43 GMT
In reply to Comment 29 (Jon):
"But is it (dragging the screens) useful anymore?"
Dragging as such isn't so useful but the double screen is still useful
for paint or image programs - the picture can completely fill the
display screen, while a control panel comes up on a second screen at
the bottom. Programs using this are Scala, DPaint, ImageMaster,
Brilliance, PageRender 3D, and others.
Conceivably these could be made to run on UAE, if AGA is ever
successfully emulated. A modern equivalent might be to use two
monitors, as we did in the early 1980s. Using multiple windows as in
Photoshop and its many imitators is a poor substitute, giving a
cluttered and confusing GUI.
Interview with Thomas Frieden: OS4 questions and answers : Comment 45 of 155ANN.lu
Posted by Anonymous on 11-Sep-2002 07:00 GMT
In reply to Comment 43 (Ben Hermans/Hyperion):
Ben - why does the use of a swap partition as opposed to a swap file make any difference to any need for a shutdown mechanism?
In both cases, either the data are already written to disk or not...
Interview with Thomas Frieden: OS4 questions and answers : Comment 46 of 155ANN.lu
Posted by Olaf Barthel on 11-Sep-2002 07:04 GMT
In reply to Comment 41 (Don Cox):
> > "Also, I wonder if OS 4 has a formal shut down procedure. If so, and
> > if Amiga users get used to the idea, no one will simply power off the
> > computer while a write is in progress."
>
> I hope it does not have a formal shutdown procedure. That is one of
> the major annoyances in other OSes.
Yes, it looks if by forcing the user to shut down the system by a well-defined
mechanism, the formal aspects finally won while the practical aspects got left
behind. I probably wouldn't mind if we had a formal shutdown procedure if,
and only if, that shutdown procedure would complete almost instantly. What I
dislike most about the approach used in today's operating systems is that it
always takes so very long to accomplish what on the Amiga is done by pushing
a button. So it takes 40 seconds to bootstrap Linux, and that's not including
the silly BIOS startup procedure. But it takes another 20 seconds to shut down
Linux on the same machine. Compare that to how older operating systems used to
handle this, e.g. NeXTSTEP on the black hardware. A complete shutdown in less
than five seconds was no exception.
But anyway, I don't believe that you can retrofit this kind of functionality
and make it a reliable feature. We have so much legacy software to support
that it will be far from practical to guarantee a well-defined response to
a shutdown command, let alone arrive with a solution that works well for
all existing software.
> Now that disk writes are so fast, there is much less chance of a file
> being partly written than in the days of 40Meg drives.
Unless, of course, application software has to shut down in an orderly fashion,
giving each instance time to dump its state information. If you cut the power
at the wrong moment, you might be interrupting something that is harder to
repair than the comparatively simple maintenance operations of a file system.
Interview with Thomas Frieden: OS4 questions and answers : Comment 47 of 155ANN.lu
Posted by Olaf Barthel on 11-Sep-2002 07:09 GMT
In reply to Comment 29 (Jon):
> But is it (dragging the screens) useful anymore? Cool maybe..Virtual
> screens might be more powerful? Matter of taste.
I believe that screen dragging is useful and has its place in the Amiga
user interface and application software design. Unlike the Apple
Macintosh and its predecessors and imitations we didn't have the one
single desktop every application would have to live on. What we did have
were applications which ran on their own virtual displays (screens), and
those screens could coexist on the same physical display. There are
certain advantages to this approach I don't want to miss. Yet, sadly, we
all had to get used to the fact that this feature wasn't always available
with the full functionality, namely screen dragging included.
I'm still rooting for getting the draggable screens back into the operating
system.
Interview with Thomas Frieden: OS4 questions and answers : Comment 48 of 155ANN.lu
Posted by Ben Hermans/Hyperion on 11-Sep-2002 07:19 GMT
In reply to Comment 47 (Olaf Barthel):
>I'm still rooting for getting the draggable screens back into the operating
>system.
You and me both.
They'll be back, just not for OS 4.0 due to a lack of time.
Interview with Thomas Frieden: OS4 questions and answers : Comment 49 of 155ANN.lu
Posted by takemehomegrandma on 11-Sep-2002 07:22 GMT
In reply to Comment 37 (Olaf Barthel):
But isn't all of this just hack and patch solutions? It doesn't feel the least lean and clean to me. How about a complete rewrite with a more modern design?
Interview with Thomas Frieden: OS4 questions and answers : Comment 50 of 155ANN.lu
Posted by Jon on 11-Sep-2002 07:22 GMT
In reply to Comment 47 (Olaf Barthel):
Yeah, indeed I don't have anything against screen dragging mechanism but this is the hands of Picasso team (right?)..I used it of course on my AGA days, but since I have been using Mediator, I haven't been able to use it, and so I have almost forgot it :(
So what are the difficulties in implementing such a feature? I was thinking that if we have screens with different resolutions, we must have some kind of scaling system, which might need a lot of CPU power and memory...then again, I don't know a lot about how Intuition or P96 work, so I may remain silent :)
Anonymous, there are 155 items in your selection [1 - 50] [51 - 100] [101 - 150] [151 - 155]
Back to Top