28-Mar-2024 11:49 GMT.
UNDER CONSTRUCTION
Anonymous, there are 18 items in your selection
[News] Francis Charig of Tao interviewedANN.lu
Posted on 07-Jan-2001 11:32 GMT by Christian Kemp18 comments
View flat
View list
Henrik Mikael Kristensen was the first to notice that CEO of TAO spills some interesting details about TAO in an amiga-news.de http://www.amiga-news.de/archiv01/010106.shtml" target="_blank">interview.
Francis Charig of Tao interviewed : Comment 1 of 18ANN.lu
Posted by Björn Hagström on 06-Jan-2001 23:00 GMT
The never ending Memory protection discussion. No MP spurs "bad programming" too in the sence of nasty viruses, memory scanners, memory trashers. Nothing would be private in a non MP system.
Aren't these reasons enough to have MP ?
Francis Charig of Tao interviewed : Comment 2 of 18ANN.lu
Posted by Hans-Joerg Frieden on 06-Jan-2001 23:00 GMT
In reply to Comment 1 (Björn Hagström):
Yes, even if security was the only argument, apart from easier debugging, program safety, virtual memory etc...
Same old sad story...
Francis Charig of Tao interviewed : Comment 3 of 18ANN.lu
Posted by Neil Thomas on 06-Jan-2001 23:00 GMT
In reply to Comment 2 (Hans-Joerg Frieden):
I agree witht the point on sloppy coding. The MP gives people an
excuse to not make their code bomb-proof. Plus the memory overhead,
the Amiga is renound for low memory overhead, perhaps it should stay
this way? Albeit a different 'Amiga' to the one we currently know.
Mobile phones etc. don't have a great deal of memory; AmigaDE must
have a small footprint.
Francis Charig of Tao interviewed : Comment 4 of 18ANN.lu
Posted by Richard Brooklyn on 06-Jan-2001 23:00 GMT
Well, I have to trow a spanner in the works and disagree, sorry.
The new AmigaDE will not need memory protection, and I am glad that we
are not getting it, in the long run it will mean better and faster
programs.
The main reason that the classic Amiga is so fast is becuase we don't
have memory protection (correct me if I'm wrong).
So what do you want? A faster computer with better programs, or a
slower computer with sloppy programs that will have more bugs
(prehaps). It's a simple trade-off: Speed verses Relibility.
Richard.
Francis Charig of Tao interviewed : Comment 5 of 18ANN.lu
Posted by Marcel on 06-Jan-2001 23:00 GMT
In reply to Comment 4 (Richard Brooklyn):
So why are you buying that argument that no mp means the DE will get better programs? The difference between my Linux box and my Amiga is that when a program craps out, the Amiga box most definitely will need a reboot, where the linux box will stay up.
MP or no mp will not determine the quality of the software. There are some situations where mp is not needed, and the overhead (very small nowadays) needs to be avoided. But there are many more situations where mp is a requirement.
How about Tao doing it the QNX way. You know, have different versions of Elate. Some with mp for devices that have the appropriate hardware to support it. Some without mp for those smaller devices that do not need to support it. I'm not touching AmigaDE without mem protection. Have enough headaches with programs on a partially mp WinME system taking the thing down.
Francis Charig of Tao interviewed : Comment 6 of 18ANN.lu
Posted by Björn Hagström on 06-Jan-2001 23:00 GMT
In reply to Comment 4 (Richard Brooklyn):
Imagine that you have logged into a remote AmigaDE-machine. At the same time
50 other users are logged in. All of them are happily reading mail, surfing
the web or whatever. You decide to do some online banking by paying some bills
from the remote AmigaDE-machine.
But what you don't know is that one of those 50 users are monitoring your
every step. Since this AmigaDE-machine doesn't have memory protection anyone
with some programming abilities can write a program that reads from any position
in memory, including your bank session. When you think that you are entering
your dear old mommys bankaccount, this user changes this number to his own
bankaccount. You happely press the send button thinking that your $200 you
borrowed last month is being payed back to mommy, while in reality they end
upp in some malicious programmers bankaccount.
Is this really what you want ? This could be something else, like someone
reading in on the loveletter you are writing to your girlfriend of whatever.
A non MP system means that there are no security what so ever. Everything
is open for everyone to read and to write.
Or one of those 50 programmers may be someone who has broken into the system
and gets the brilliant idea to write a simple program that fills the whole
memory with zeros. Oups, the system crashed, throwing out everyone that was
logged in, destroying whatever they was working on. Had there been MP, the
process that tried to fill memory would simply had been terminated and no
one else wouldn't even have noticed anything.
The future is not one person systems like the classic AmigaOS is. The future
is multiuser systems, and it has been for a long time. Just look at the
different *nix'es out there.
Will people please step out into reality ?
Francis Charig of Tao interviewed : Comment 7 of 18ANN.lu
Posted by Hans-Joerg Frieden on 07-Jan-2001 23:00 GMT
In reply to Comment 3 (Neil Thomas):
I always find this argument about sloppy programming just a nice excuse for not having MP. Defective programs will *always* bomb, either with or without, and I think it more than a little strange to assume that its absence will make programs better.
The up side of MP is that it actually has a discernable effect when the program bombs. You don't have to go guessing when it bombs.
I can't stand this whole talk about "lean OS". A few more kilobytes will not hurt anyone, too many times this "lean OS" argument has been used as an excuse for the lack of technical progress.
Best regards, Hans-Joerg.
Francis Charig of Tao interviewed : Comment 8 of 18ANN.lu
Posted by the man in the shadows on 07-Jan-2001 23:00 GMT
In reply to Comment 6 (Björn Hagström):
> Imagine that you have logged into a remote AmigaDE-machine. At the same
> time 50 other users are logged in. All of them are happily reading mail,
> surfing the web or whatever. You decide to do some online banking by
> paying some bills from the remote AmigaDE-machine.
> But what you don't know is that one of those 50 users are monitoring your
> every step. Since this AmigaDE-machine doesn't have memory protection anyone
> with some programming abilities can write a program that reads from any position
> in memory, including your bank session. When you think that you are entering
> your dear old mommys bankaccount, this user changes this number to his own
> bankaccount. You happely press the send button thinking that your $200 you
> borrowed last month is being payed back to mommy, while in reality they end
> upp in some malicious programmers bankaccount.
This proves that you do not know very much about memory protection. However what you do know is how to point out bad programming. If the person that created the server has done this type of programming to allow someone to break through to the server and run a program that "mirrors" the ports then that is the fault of program that is running the server... not memory protection. The same water holds true if the hacker wants to break in and zero out the memory. You can do both with _and_ without memory protection even in NetBSD if you program it right and NetBSD is probably the most famous for having one of the best security systems around.
I've said it before and I will say it again... lack of memory protection does not mean that a good program will crash in flames, it only supports good programming a causes the programmer to check and recheck the program until it works properly. I don't want memory protection on my system as it will promote nothing but sloppy half-ass slap together bloatware that is really uncalled for in the first place. This is the very reason why many Amiga users use programs like Enforcer, Sushi, and other reporting tools of the like.
Francis Charig of Tao interviewed : Comment 9 of 18ANN.lu
Posted by Anonymous on 07-Jan-2001 23:00 GMT
In reply to Comment 8 (the man in the shadows):
This shows you know nothing about memory protection, actually.
Without memory protection from the OS, any program can read and write any other programs memory. Even if the programs themselves have different address spaces, these are all mapped by the mmu into the same address space. So writing a program that randomly writes rubbish into memory is a simple way of killing a computer entirely, as writes outside of the allowed memory space for a process are not caught, so these writes will affect another area of memory which could be in use by another program.
So you are running a web server for 100's of sites, with many users logged in at the same time. One user runs some software that does something wrong, and the whole system crashes and needs rebooting. Four hundred miles away. Great one whoever decided that was a good idea.
I do not know if TAO have an alternative system for preventing memory accesses outside of a processes address space, but I doubt it. The benefits of memory protection outweight the disadvantages by several orders of magnitude in my opinion. There is no excuse for not having MP as a module within Elate. If the computer has the required hardware, then use it, else don't use it - it shouldn't affect the software running on it except that on some machines a nasty program that messes the system up will crash the machine entirely whereas on another machine a window will pop up saying "Program X tried to access memory it shouldn't have and has been closed".
Even the most embedded of CPUs in the future will have an MMU anyway.
Also, my wife has a phone that occasionally crashes.
Francis Charig of Tao interviewed : Comment 10 of 18ANN.lu
Posted by Anonymous on 07-Jan-2001 23:00 GMT
In reply to Comment 9 (Anonymous):
The only way that I can see that Elate can protect memory is because of the Elate Assembler which stuff is written in, and compiled upon loading by the OS. When the code is being loaded and assembled, the compilar could probably catch a lot of the errors in the code that would have caused a memory access error normally. Of course, this relies on all code being distributed in the Elate VC format, and not in native formats, and it is impossible to detect whether a memory access to location X + y would be within the programs memory space or outside unless this is all being done at runtime - a huge overhead as all memory accesses would invoke an OS exception to check that it was valid - fine for debugging, not so good for running real programs!
The lack of any form of virtual memory is also worrying. Sure, paging memory on a windows machine is nasty, but you hardly notice it on a good operating system, and it enables far more features than the disadvantages that it creates (oh no, this program is taking more than 1 second to do something). I know that Elate is a RTOS, and that this greatly affects guarantees of real time behaviour, but for a consumer operating system it is essential...
Francis Charig of Tao interviewed : Comment 11 of 18ANN.lu
Posted by the man in the shadows on 07-Jan-2001 23:00 GMT
In reply to Comment 9 (Anonymous):
> Also, my wife has a phone that occasionally crashes.
I have a windows2000 system that crashes as well. What's your point?
Francis Charig of Tao interviewed : Comment 12 of 18ANN.lu
Posted by Anonymous on 07-Jan-2001 23:00 GMT
In reply to Comment 11 (the man in the shadows):
In the interview, the bloke from TAO says that mobile phones don't crash, even though they don't have MP. Yet that phone crashes. Not looking good for non-MP systems eh?
When Program X goes down, I don't want it taking out programs Y and Z and the whole damned OS...
Francis Charig of Tao interviewed : Comment 13 of 18ANN.lu
Posted by Leif on 07-Jan-2001 23:00 GMT
There should be some differences between
different memoryprotection technics, right ?
some slower but more secure, others faster and more tolerant.
surely som kind of it would fit in there.
Personally, I would like MP when developing mostly.
Francis Charig of Tao interviewed : Comment 14 of 18ANN.lu
Posted by the man in the shadows on 07-Jan-2001 23:00 GMT
In reply to Comment 12 (Anonymous):
> Not looking good for non-MP systems eh?
You've completely missed the point. Bad programming, bad crashes... good programming, no crashes... MP or not. Take a look at the TI-99 calcs... can they crash? Do they have memory protection? What about the Psion. Do they crash? Well, I can vouch for the Psion systems as yes, they can crash... but it took me about four days to get the damn thing to crash with nothing in mind but to get it to crash, that alone says something right there. I also know that the Psion systems do not have any form of memory protection. Windows2000 has memory protection... I can bring that down in about 30 seconds after booting. Memory protection means absolutely nothing to me except for lazy programming tactics.
Not looking good for your argument eh?
Francis Charig of Tao interviewed : Comment 15 of 18ANN.lu
Posted by Roj on 07-Jan-2001 23:00 GMT
Okay, let's think about this "memory protectionvirus protection" idea...
Microsoft Windows has more viri circulating, resulting in more virus protection software written for it than any other platform in existence.
It also has memory protection. I guess the Norton and McAfee fellas don't know much about memory protection either eh?
Truth be known, with or without memory protection, a malicious virus could wipe a hard drive clean, propogate itself through the network and that would be that. Thankfully that's not happened yet, but it could, and there's not a thing memory protection could do to stop it.
Now, what real merits does memory protection bring to the table?
Francis Charig of Tao interviewed : Comment 16 of 18ANN.lu
Posted by Björn Hagström on 07-Jan-2001 23:00 GMT
In reply to Comment 15 (Roj):
Okay, let's think about this "memory protectionvirus protection" idea...
Microsoft Windows has more viri circulating, resulting in more virus protection software written for
"it than any other platform in existence.
It also has memory protection. I guess the Norton and McAfee fellas don't know much about
memory protection either eh?
Truth be known, with or without memory protection, a malicious virus could wipe a hard drive
clean, propogate itself through the network and that would be that. Thankfully that's not happened
yet, but it could, and there's not a thing memory protection could do to stop it.
Now, what real merits does memory protection bring to the table?"
No one has stated that MP is the ultimate virus protection. Although, MP makes
is harder for a virus to be really mean since it would only have access to
what the userspace it is executed in has access too. In a non MP system it
would have access to exacly everything in a system independently to the
userspace.
Francis Charig of Tao interviewed : Comment 17 of 18ANN.lu
Posted by Björn Hagström on 07-Jan-2001 23:00 GMT
In reply to Comment 8 (the man in the shadows):
>This proves that you do not know very much about memory protection.
I'm sorry, but I happen to know quite alot about memory protection. I'm
very sorry if this disturbs you.
>However what you do know is how to point out bad programming. If the
>person that created the server has done this type of programming to
>allow someone to break through to the server and run a program that
>"mirrors" the ports then that is the fault of program that is running
>the server... not memory protection.
Sorry again, you totally missed the idea. I was not talking about
mirroring ports. I was talking about scanning non protected memory.
We are talking MP here remember?
>The same water holds true if the hacker wants to break in and zero
>out the memory. You can do both with _and_ without memory protection
>even in NetBSD if you program it right and NetBSD is probably the most
>famous for having one of the best security systems around.
In a MP system you would need very high access to do such a thing. In
a non MP system any level of access would suffice since everyone has
access to everything.
>I've said it before and I will say it again... lack of memory protection
>does not mean that a good program will crash in flames, it only supports
>good programming a causes the programmer to check and recheck the program
>until it works properly.
I'm guessing to what you mean here. But if you are stating that programs
are automatically better if they were made in a non MP system then you
are totaly wrong. MP would rather help the programmer than vice versa.
>I don't want memory protection on my system as it will promote nothing
>but sloppy half-ass slap together bloatware that is really uncalled for
>in the first place. This is the very reason why many Amiga users use
>programs like Enforcer, Sushi, and other reporting tools of the like.
Im sorry again, memory protection has nothing to do with sloppy half-ass
slap together bloatware. Only lazy programmers has to do with that.
People are using programs like Enforcer, Sushi and other reporting tools
just because of the lack of MP and resourcetracking i AmigaOS. And they
are constantly swearing over the lack of them.
Francis Charig of Tao interviewed : Comment 18 of 18ANN.lu
Posted by Björn Hagström on 07-Jan-2001 23:00 GMT
In reply to Comment 14 (the man in the shadows):
>You've completely missed the point. Bad programming, bad crashes...
>good programming, no crashes... MP or not.
And you have missed the point of MP. Without MP, the whole system is
Anonymous, there are 18 items in your selection
Back to Top