[9fans] A prick into the wasps' nest ;-)

[9fans] A prick into the wasps' nest ;-)

[cej]

Hi, friends,

I post, with permission, an excerpt from a debate between me and Wlad Los wlad@mipk.kharkiv.edu ...
... would love to hear reactions from you!

Best regards,
++pac.

===============
> Vladimir Los wlad@mipk.kharkiv.edu wrote:
> 
> http://www.oberon.ethz.ch/native/WebScreen.html
> :o)
> And it was the same interface in the middle of 80ths...
> As always IT community doesnt think about Wirth's work seriously. And
> then (after years) makes similar solutions...
> 

++pac:
Perhaps you don't know that Acme was indeed inspired by Oberon interface, see http://cm.bell-labs.com/sys/doc/acme/acme.html
however, it seems to me [without personal experience with it] that Oberon's menus are ordinary menus, not ordinary *text* that can be inserted/modified by the user...

Wlad:
> Oh! excuse me! Really I didn't know about it!
> 
> > however, it seems to me [without personal experience with it] that
> > Oberon's menus are ordinary menus, not ordinary *text* that can be 
> > inserted/modified by the user...
> 
> Now it is my turn to say "Perhaps you don't know that..." :o)
> In Oberon there is nothing but OBJECTS! You can insert/modify 
> EVERYTHING. There is no difference between any things in the system 
> ("files", texts, pics, labels, text formatters and scanners, items of 
> menu - as you asked, ...) everything can be accessed and be 
> manipulated 
> by the same way. And it was from the beginning of Oberon system 
> (aproxim. 1986 - I think...)
> You can insert different controls in "text" or you can insert 
> text into 
> a "form" or you can even compile a button label (if you want 
> think about 
> it as a program text) :o)))))
> Sometime I wonder very much why programmers doesn't know about these 
> systems. May be programmers when they hear word "Wirth" think 
> "Ahhh, I 
> know Wirth is a Pascal inventor! But Pascal is the language for 
> educational goals and not for "real" and "serious" programming. 
> Consequently the systems developed by Wirth can not be serious." :o)))
> 
> But I sometime think that Wirth "took the baton from
> bell-labs" when he 
> return from there. Almost all famous solutions were made in 
> Zurich and 
> then applied by "grands" of IT industry...
> 

very interesting, but how can I live without C?

:o))))
There were my thoghuts some time ago too!!! Of course I have to spend 
some time to teach new system - but you have to do this for every new 
system! :o)
And yet one thing! Oberon diffirintiate from other world very much! It 
is not "better" or "worse" - it is DIFFERENT. I have to take a habit... ;o)

> I'm a biologist and I need to run also ordinary "programs"
> I have ported 50+ ones from various dialects of C to plan9 native C, 
> and now I should learn Oberon language and port... :-(

I am not a "violator". I have some spare/free time and I have seen your 
message... ;o)
Of course it is your decission to learn thomething or not... ;o) I just give you an advice. And my impressions. I know c/c++, Modula-2, 
Forth, Delphi, Smalltalk. The better it willl be to say "I worked in 
them". But three years one my friend shown me Aos (it is the last of 
ETHZ OS) And I will be "killed"... :o) THeir model of multitasking (and 
synchronization) is something!!! Aos (or bluebottle 
http://www.bluebottle.ethz.ch) is the result of 20 years researches. I 
think it is the same sutiation - NOBODY else couldn't do the same OS. 
And community even doesn't realize and understand WHAT has appeared!!! 
;o) The most of programmers are under the presse of old paradigms? 
models and thinkings...

++pac:
 > In Oberon, can I just rewrite the text on a button of a running 
program, without recompilation,
 > and the program just does this another command?
 > Acme does this.

By this example I tried to show the flexibility of the system... :o) But! Of course YOU CAN CHANGE THE behavior of every control without 
recompilation. There many ways to do this. One of them - to change value 
of field Link. :o)

And yet some words...

Rob Pike rob@plan9.bell-labs.com
Originally appeared in Proc. of the Winter 1994 USENIX Conf., pp. 
223-234, San Francisco, CA

"Cedar was, however, the major inspiration for Oberon [Wirt89], a system 
of similar scope but much smaller scale. Through careful selection of 
Cedar's ideas, Oberon shows that its lessons can be applied to a small, 
coherent system that can run efficiently on modest hardware. In fact, 
Oberon probably errs too far towards simplicity: a single-process system 
with weak networking, it seems an architectural throwback."

Sometime I think why creators of most famouse OS of XX cent. look like 
children or like liars. If you can remember other "father" of Unix Brian 
Kernighan in his Why "Pascal is Not My Favorite Programming Language" 
criticizes Pascal.

But in both cases their "arguments" and "facts" are very week and may be 
even false! It concernrs Brian's knowing of pascal development and Rob's 
estimation of Oberon...

One interesting question is Why do they speak wrong things? May be they 
can look Oberon's potentials and they try to "avert" the attention of 
community from Wirth's systems and works. But then they use Wirth's 
result in their works! This is very interesting! I look at Limbo and 
what I can see - Oberon with a mask! All the time they say "garbage 
collection is a very bad thing", but introduce it into Inferno. "Byte 
code is a waste of time" - but use it after Sun (which in their turn 
used GC and byte-code after slim-binaries of Oberon).

Of course Oberon (and its desc-s) are not so popular. But we have veru 
strange state in our branch. Everybody use "popular" systems. These 
systems use solutions developed in Wirth's labs. But these solutions in 
those popular systems are not "native". They are fasten to those systems 
with restrictions and under "other ideology" - they can not reveal all 
of their advantages... In result we have half-decisions and we have to 
correct problems due to them...

And get looped circle again!


Wlad:
Now I am reading and I think that I walked in the dark forest using else 
but not Oberon and Plan9.
Now I am even afraid I have to go to Plan9! :o)))
I like it more and more while I'm reading about it! :o)
I'm going to stay after my work today and download it...


++pac:
> [Oeron] has things
> that Plan9 does not, eg. objects -- I suppose that GUI in Oberon is just a
> dynamic object that other programs-objects talk to,

Something different. GUI IS a set of objects. We have several defined 
interfaces and you can implement them and achieve neded functionality. 
Of course the system has predefined implementations of interfaces and 
then it works... :o)

> so any change to this single object affects all other programs talking 
> to it, am I right?

Yes as we could see if you change for example any property (through 
Colombo - program-object to access working objects) this change 
transforms the object "on the fly". More You can change implementation 
of any objects without "stopping" the system or any part of it. For example I just can relink refference to other "window manager" and 
my interface becomes 3D or like in Windows...

Of course to achieve such flexibility you (your objects) must fulfil 
some rules (support defined interfaces).

And the most pleasure for me is GARBAGE COLLECTOR!!!!!!!!
I know I am not the most attentive homo in this Universe. I make a great 
deal of mistakes concerning alloc/free of memory. May be it is due to my 
thoughts about rules of proper language are flying behind the thoughts 
about problem I am solving... :o)
But with GC I became to focus only on problem domain not on system low 
level specific things.

I have got very strong habit to program in GC-language. IT'S VERY 
COMFORTABLE AND YOUR PRODICTIVITY GROWS MUCH MORE! And I am not going to 
change such class of systems to something else. May be Java (and system 
based on it), or Limbo (in Inferno), or Smalltalk... :o)

> Plan 9 even has no shared libs (like .so, .dll), thus you have to 
> relink everything each time you make a change in a library. Everything 
> is statically linked. They state that shared libs are not needed, 
> though not harmful if appropriately used.

???????????????????????????
It is something new for me!!!! And Pike can write that Oberon is step 
backward????????? :o))))
The dynamic linking and loading is one of the "elephant" to achieve 
flexibility.

Now ETHZ is ending new core of Aos. It becomes real-time. That's great news!

> nevertheless, shared libs are not true objects. I'm not the proper 
> person to discuss pros and cons of OO paradigm. I was pursuaded I can 
> live w/o objects :-)

Oh, no! It's just a mirage! I can assure you 101% :o) I can't live w/o 
objects! Nobody can! :o) Unix was so successful because of it was 
defined one very useful and common OBJECT - "FILE"... :o)))) And I wont 
speak about graphical library. May be you can't inherit from library's 
object but the ideology ot those libraries is object.

===============

Re: [9fans] A prick into the wasps' nest ;-)

[Rob Pike]

acme and the acme paper were written many years ago,
when oberon was vastly less capable than it is now.
oberon was and is an interesting system; you know
what they say about imitation.

however, even today, i don't think oberon has an analog
to the key new idea in acme: acquisition with button 3.

-rob

Re: [9fans] A prick into the wasps' nest ;-)

[Vladimir Los]

Rob Pike wrote:

> acme and the acme paper were written many years ago,
> when oberon was vastly less capable than it is now.
> oberon was and is an interesting system; you know
> what they say about imitation.

Excme, Rob, what do they say about imitation? :o)

> however, even today, i don't think oberon has an analog
> to the key new idea in acme: acquisition with button 3.

What's wrong with button 3?

Have you any plans to add real-time features to Plan9 or Inferno?
Where can I read about requirements to resources for minimum 
installation of Inferno?
Did you heared about QNX? What is your oppinion about it?

[9fans] wasp's answer... :o)

[Vladimir Los]

This is the answer from one of the participant of Oberon discussion list
http://www.delphikingdom.com/asp/talktopic.asp?ID=285&ref=msg&msg=1402#msg1402 
. The text is in cyrillic.

[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[
На ?сообщение 1401? (Владимир):
Прочитал статью на http://www.citforum.ru/operating_systems/inferno/. 
Первая часть - полностью описание оберон-системы. Вторая часть - 
описание кастрации оберон-системы под соусом Великих Усовершенствований. 
Это что, шутка?
]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]]

My translation of it:

[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[
On message 1401 (Vladimir)
I have read an article at http://www.citforum.ru/operating_systems/inferno/.
First part is an Oberon system abstract desription. Second part is a 
description of a castration (castrating) of Oberon system with a 
dressing (sauce) "Great Improvements".
Was it a joke?
[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[[

Re: [9fans] A prick into the wasps' nest ;-)

[Sape Mullender]

> Have you any plans to add real-time features to Plan9 or Inferno?
> Where can I read about requirements to resources for minimum 
> installation of Inferno?

man proc

RE: [9fans] A prick into the wasps' nest ;-)

[Mark F Rodriguez]

> > Have you any plans to add real-time features to Plan9 or Inferno?
> > Where can I read about requirements to resources for minimum
> > installation of Inferno?
> 
> man proc

Probably more suited for the Inferno list, but man proc results in "No
matches" when issued from the Windows-hosted emu. I read the man page online
and it would be very nice to have a similar utility in Inferno. Am I missing
something?

Thanks,
-Mark

Re: [9fans] A prick into the wasps' nest ;-)

[William Josephson]

On Tue, Jun 08, 2004 at 02:55:29PM -0400, Mark F Rodriguez wrote:
> Probably more suited for the Inferno list, but man proc results in "No
> matches" when issued from the Windows-hosted emu. I read the man page online
> and it would be very nice to have a similar utility in Inferno. Am I missing
> something?

Try it under Plan 9 or read the Plan 9 manual pages on the web.

RE: [9fans] A prick into the wasps' nest ;-)

[rog]

> Probably more suited for the Inferno list, but man proc results in "No
> matches" when issued from the Windows-hosted emu. I read the man page online
> and it would be very nice to have a similar utility in Inferno. Am I missing
> something?

inferno does not currently have any realtime support. (it's not
clear to what extent this could be done on a non-native
platform anyway).

also, its process device is called "prog", not "proc", so "man prog"
or "wm/man prog" gives a more useful result.

RE: [9fans] A prick into the wasps' nest ;-)

[Mark F Rodriguez]

> inferno does not currently have any realtime support. (it's not
> clear to what extent this could be done on a non-native
> platform anyway).

I'm not sure about other platforms, but don't Linux(2.6), FreeBSD Plan9
(from reading man proc), and Windows support [soft] realtime within the OS
out-of-the-box? I know you can get a POSIX 1003.1b implementation to bring
your OS (at least those with source available) up to realtime specs. Perhaps
one could [re]build the platform emu with a flag setting specifying whether
or not they have realtime support (Windows, Plan9 or POSIX)???

I can easily see how this could present a challenge on non-native platforms
even if the hosting platform had real-time support, but I can also see great
benefits to having realtime capabilities within native Inferno. Are there no
plans to support proc (perhaps with different results depending on host w/o
realtime support vs. native Inferno or host with realtime support)?

> 
> also, its process device is called "prog", not "proc", so "man prog"
> or "wm/man prog" gives a more useful result.

Thanks for the clarification, I had recently read man prog but was hoping
the realtime features of proc (i.e., period, deadline, cost, yieldonblock,
etc.) where somewhere hidden within Inferno.

Thanks,
-Mark

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> Thanks for the clarification, I had recently read man prog but was hoping
> the realtime features of proc (i.e., period, deadline, cost, yieldonblock,
> etc.) where somewhere hidden within Inferno.

does embedded/dedicated system mean anything?

Re: [9fans] A prick into the wasps' nest ;-)

[vdharani]

Bruce,

> $50 386 from china.  copy the conf file and delete
> the drivers you don't need.  type mk.  enjoy.
can you provide any pointers?

> or the wrt54g router is a more expensive minimum.
i want to buy one and try it out. will you or someone be able to provide
me instructions on how to get started? Can I do tftp boot?

how easy is it to get started with this hardware?

for a long time, i wanted to play with this. if someone can help me, it
would be great.

thanks
dharani

[9fans] A prick into the wasps' nest ;-)

[cej]

> acme and the acme paper were written many years ago,
> when oberon was vastly less capable than it is now.
> oberon was and is an interesting system; you know
> what they say about imitation.

dunno?

> however, even today, i don't think oberon has an analog
> to the key new idea in acme: acquisition with button 3.

I'm not going to switch to oberon ;-))
i think it doesn't have anything like buton3 acquisition,
and piping to dot. And plumbing, of course.
BTW, old Oberon is dead, and Bluebottle has much more
widozish look'n'feel, which i don't like...
Not to say, how could I live without C?

However, some (maybe) provocative  questions remain, namely :
is OO paradigm overcome/superseeded/abandoned, or has more cons than pros?

Re: [9fans] A prick into the wasps' nest ;-)

[Vladimir Los]

Good Day, Peter!

I am here at last... :o)

> I'm not going to switch to oberon ;-))
> i think it doesn't have anything like buton3 acquisition,

What do you and Rob talk about every time? Is button 3 role something 
special in our life? :o))) Oberon not only use all three buttons, but 
inserts so called "interclicks"... Due to them the count of functions, 
which can be done by mouse (without keyboard shortcuts) increases...

> and piping to dot. And plumbing, of course.

I've started to learn Plan 9 as you know three days ago. What do you 
mean saing about these things? Is it analog of "clipboard"? In Oberon 
you can manipulate selected chunks of texts (not only texts but objects 
of any types!). Of course Oberon has no analog of pipe (yet, because of 
needless), but you can refer to need info by using special sintax 
agreements (*, ~, ^).
Additionally, as I can see what Plan9/Inferno users can do only in acme, 
Oberon users can do everywhere in the system. Any part (chaunk) of text 
fulfiled syntax agreement (Module.Procedure) can be executed as "command"...

> BTW, old Oberon is dead,

A few brave sentence. They just transformed to new generation systems.

  and Bluebottle has much more
> widozish look'n'feel, which i don't like...

Again. You have to see a great influance of Windows to "masses". Oberon 
had very reach mouse actions combinations. But mainstream prefer 
"windozish" approaches. I do not like Windows too. But I can say "yes 
natural selection" (dirested by one firm - you know what I mean :o) ) 
made its work: nobody use Oberon or Inferno. Almost everyone use Windows 
and has habit to its mouse actions agreement...

Of course we can create systems which are enough flexible to setup by a 
user for his/her habits... But where are such systems? Linux?

> Not to say, how could I live without C?

As I can see Inferno users can! :o)
And I agree with them. More! If I work in Unix-like sustems, I prefer 
Inferno just because it has many features looking like Oneron's ones. :o)

> However, some (maybe) provocative  questions remain, namely :
> is OO paradigm overcome/superseeded/abandoned, or has more cons than pros?

Odf course I see this question is addressed to Rob but may I answer it too?
OO paradigm is not the question of programming (coding if you want). 
This is the thing of thinking and design. Unix became so popular due to 
its authors (designers) made OO decision: they started to think about 
everything like about files. Thinking such way they have restricted a 
domain of solutions, but at the same time they have desciplinized the 
thinking of great deal of programmers...
If you think you can now not to take into account OO-design principles 
just because last OSs were programmed and look like "nonOO", you are 
wrong. OO can help you to create more structured and accurate programs. 
You can start thinking more productively when you "prepare" you domains. 
You can start undersatnd you program texts after several months after a 
work is done. :o)

Re: [9fans] A prick into the wasps' nest ;-)

[kim kubik]

----- Original Message -----
From: cej@gli.cas.cz <cej@gli.cas.cz>
> > is OO paradigm overcome/superseeded/abandoned, or has more cons than
pros?
------
From: Vladimir Los <wlad@mipk.kharkiv.edu>
>  may I answer it too?
> OO paradigm is not the question of programming (coding if you want).
> This is the thing of thinking and design.
-------
This interview might be of interest (she's a Sun "Java Evangelist"):

  "... we seem to have reached the point where OO is no longer effective.
No one can comfortably negotiate a system with thousands of classes. So,
unfortunately, object-oriented programming has a fundamental flaw,
ironically related to its main strength."
    - excerpted:
  The Next Move in Programming: A Conversation with Sun's Victoria
Livschitz,
   -- Senior IT Architect and Java Evangelist, Sun Microsystems

http://java.sun.com/developer/technicalArticles/Interviews/livschitz_qa.html

She also says:
  "... An enormous mess of XML documents that are now being
created by enterprises at an alarming rate will be haunting our industry for
decades."

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

>   "... we seem to have reached the point where OO is no longer effective.
> No one can comfortably negotiate a system with thousands of classes. So,
> unfortunately, object-oriented programming has a fundamental flaw,
> ironically related to its main strength."
>     - excerpted:
>   The Next Move in Programming: A Conversation with Sun's Victoria
> Livschitz,
>    -- Senior IT Architect and Java Evangelist, Sun Microsystems

nuthin' wrong with OO.  all decent C programmers write in an OO
style.  what IS wrong with it is that the core of these OO languages
[java, python, C#, ...] is rotten.

i don't need no thousands of classes:

    http://minnie.tuhs.org/UnixTree/Nsys/sys/nsys

Re: [9fans] A prick into the wasps' nest ;-)

[George Michaelson]

from kerneltraffic.org...

	This was one of those, "this would be weird and great", "no it wouldn't"
discussions. Sergiy Lozovsky started it up innocently enough, asking how
to deal with kernel stack overflow. Then it turned out that "I put LISP
interpreter inside the Kernel - http://vxe.quercitron.com - It works, but
it use a lot of stack memory. It's impossible to rewrite it easily, though
I'll investigate why exactly it uses so much of stack memory" .

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> This was one of those, "this would be weird and great", "no it wouldn't"
> discussions. Sergiy Lozovsky started it up innocently enough, asking how
> to deal with kernel stack overflow.

yeah, you take a kernel segment fault.  that pretty much finishes up
operations early, so you can do a lotta surfing.

Re: [9fans] A prick into the wasps' nest ;-)

[andrey mirtchovski]

> http://minnie.tuhs.org/UnixTree/Nsys/sys/nsys

ah, here's where the name 'rc' originated!

http://minnie.tuhs.org/UnixTree/Nsys/sys/nsys/ken/rc.html

from the commentary:

	Even the compilation mechanism is a bit unclear, though it
	certainly used the shell script ken/rc, which appears to
	depend on having the *.o files from ken/* and dmr/* and also
	slib.a, which doesn't exist on the tape.

i love history :)

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> ah, here's where the name 'rc' originated!
> 
> http://minnie.tuhs.org/UnixTree/Nsys/sys/nsys/ken/rc.html

rc?  runcom -- run commands,  right?

Re: [9fans] A prick into the wasps' nest ;-)

[Geoff Collyer]

Right, rc is an abbreviation of runcom, which in turn comes from CTSS.

http://www.faqs.org/faqs/unix-faq/faq/part1/section-3.html
has a longer explanation.

Re: [9fans] A prick into the wasps' nest ;-)

[rog]

> "... we seem to have reached the point where OO is no longer effective.
> No one can comfortably negotiate a system with thousands of classes."

this is really due to inheritance. such a handicap.

Re: [9fans] A prick into the wasps' nest ;-)

[Vladimir Los]

kim kubik wrote:
> -------
> This interview might be of interest (she's a Sun "Java Evangelist"):
> 
>   "... we seem to have reached the point where OO is no longer effective.
> No one can comfortably negotiate a system with thousands of classes. So,
> unfortunately, object-oriented programming has a fundamental flaw,
> ironically related to its main strength."
>     - excerpted:
>   The Next Move in Programming: A Conversation with Sun's Victoria
> Livschitz,
>    -- Senior IT Architect and Java Evangelist, Sun Microsystems
> 
> http://java.sun.com/developer/technicalArticles/Interviews/livschitz_qa.html
> 
> She also says:
>   "... An enormous mess of XML documents that are now being
> created by enterprises at an alarming rate will be haunting our industry for
> decades."
She (and somebody else) can say more and more but if we continue to have 
the main thing for reusing of code through inheritance of classes we 
continue to have "major flaws of OO".

RE: [9fans] A prick into the wasps' nest ;-)

[Mark F Rodriguez]

> adding sape's edf code to inferno takes an hour.

I just read the paper "Lightweight EDF Scheduling with Deadline
Inheritance" and it was an interesting read. Has someone already done this
for Inferno or were you talking theoretically?

Thanks,
-Mark

Re: [9fans] A prick into the wasps' nest ;-)

[Bruce Ellis]

i did it with an earlier version of the edf code.

Mark F Rodriguez wrote:
>>adding sape's edf code to inferno takes an hour.
> 
> I just read the paper "Lightweight EDF Scheduling with Deadline
> Inheritance" and it was an interesting read. Has someone already done this
> for Inferno or were you talking theoretically?
> 
> Thanks,
> -Mark

Re: [9fans] A prick into the wasps' nest ;-)

[rog]

> i did it with an earlier version of the edf code.

was this with dis progs, or just with kprocs?

Re: [9fans] A prick into the wasps' nest ;-)

[Bruce Ellis]

kprocs.

rog@vitanuova.com wrote:

>>i did it with an earlier version of the edf code.
> 
> was this with dis progs, or just with kprocs?

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> > adding sape's edf code to inferno takes an hour.
> 
> I just read the paper "Lightweight EDF Scheduling with Deadline
> Inheritance" and it was an interesting read. Has someone already done this
> for Inferno or were you talking theoretically?

if brucee says it takes an hour, it takes an hour.

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> Practitioners in the field of real-time computer system design often
> distinguish between `hard' and `soft' real-time systems.  Hard real-time
> systems are those where it is absolutely imperative that responses occur
> within the specified deadline.  Soft real-time systems are those where response
> times are important but the system will still function correctly if deadlines
> are occasionally missed.
> ...
> the softness isn't a licence for sloppiness.

well put.

RE: [9fans] A prick into the wasps' nest ;-)

[vdharani]

charles,

> Of course, it's much the same as Dharani's definition except that the
> `soft' example is slightly better for me:
> i didn't like the idea of missing frames when DVD playing!  reduced value
> indeed!
>
> the softness isn't a licence for sloppiness.

i am wondering if the wide popularity of DVD players followed by the
quality improvement makes you think so.

thanks
dharani

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> i am wondering if the wide popularity of DVD players followed by the
> quality improvement makes you think so.

humans can cope with with a a few missed frames.  there are many 
factors involved.  if we assume that TV is 26 (iirc) frames/second
there is phosphor persistance on the the screen (let's forget about
LCD's) and (iirc) there's retinal persistance.  so, dropping a
few frames  doesn't really matter.  oh yeah, it's interleaved as
well.

they may not even be watching when the frames are dropped.

if yer talking about military or medical systems they have to
be 'right on time' within certain tolerances.  otherwise yer
F-16 turns into an unflyable mess or your patient dies.

if you wanna build a 'real time' system:

    1. define it
    2. design it
    3. code it
    4. test it till it breaks and should it, goto 1

otherwise, pick up a hammer and go search for some (soft) approximation of a nail's position.

btw: as Seigfried said "WE DON'T DO THAT AT KAOS"

RE: [9fans] A prick into the wasps' nest ;-)

[Nigel Roles]

> factors involved.  if we assume that TV is 26 (iirc) frames/second
> there is phosphor persistance on the the screen (let's forget about
> LCD's) and (iirc) there's retinal persistance.  so, dropping a
> few frames  doesn't really matter.  oh yeah, it's interleaved as well.
> 
> they may not even be watching when the frames are dropped.
> 

Even if I softly drop a few words here and there, this is still
recognisably rubbish.

The frame rate is 25Hz for PAL, or 30Hz for NTSC (29.97 for pedants).
SECAM is also 25 Hz.

There is some persistance (sic), but the human eye/brain
is extremely sensitive to missing frames, or disturbed frame timing,
so it is arguable that persistence makes a timing error worse.
You only have to watch a someone walk across the shot on a decent
DVD player and then compare it will your typical soft decoder on
a Windows laptop to know that seeing all the fields at the right
time is a really important.

Interleaving only helps reduce apparent flicker. Each field represents
a different point in time, so missing a field, or replacing one
with a copy of the previous one is ruinous. Swap the fields and
everything looks furry. 100Hz TVs don't just
display each field twice; they have a DSP in there to
sort out the motion compensation and synthesise the missing fields.
My 100Hz TV still screws up on fast scrolling credits (horizontal
and vertical).

Stick to what you know, Boyd.
Except the guns; it's really, really tedious.

RE: [9fans] A prick into the wasps' nest ;-)

[Steve Simon]

> There is some persistance (sic), but the human eye/brain
> is extremely sensitive to missing frames, or disturbed frame timing,
> so it is arguable that persistence makes a timing error worse.

Perhaps I am a pedant, but the missing frames are not the problem
but the disjointed motion is.

The human perceptual system has evolved to be good at predicting
where moving objects will go (eg cricket balls) we can even
"understand" second order changes (cars accelerating) but high
order or disjoint motion looks "wrong". This is a major
headache when designing motion estimators - people are very good
as noticing where it goes wrong (like your TV). 

-Steve

Re: [9fans] A prick into the wasps' nest ;-)

[Bruce Ellis]

i get really annoyed with badly mastered dvds, let alone
crap players that are worse than VHS.  i get agitated by
a dropped frame, particularly an I frame, and will return
the DVD as faulty if it happens in the "good bits".  yeah
i have 100Hz component which does more video formats than
i have the ITUs for.  even worse are the DVDs that slip
lip-sync 'cause they are badly mastered.  when the lips
and the voice get 2 frames apart i give up.  remarkable
that i can master a DVD-R from DV - at a high bit rate and
get no such problems.  this has little to do with real-time,
but ... the reason i added edf to the ps2 inferno kernel
was so the double buffering was in sync with the retrace
and the IDCT dmas were in time for the buffer.  i think
the linux guys are still working on it and are happy with
5 fps, 'cause it's a start ...

brucee

Nigel Roles wrote:
>>factors involved.  if we assume that TV is 26 (iirc) frames/second
>>there is phosphor persistance on the the screen (let's forget about
>>LCD's) and (iirc) there's retinal persistance.  so, dropping a
>>few frames  doesn't really matter.  oh yeah, it's interleaved as well.
>>
>>they may not even be watching when the frames are dropped.
>>
> 
> 
> Even if I softly drop a few words here and there, this is still
> recognisably rubbish.
> 
> The frame rate is 25Hz for PAL, or 30Hz for NTSC (29.97 for pedants).
> SECAM is also 25 Hz.
> 
> There is some persistance (sic), but the human eye/brain
> is extremely sensitive to missing frames, or disturbed frame timing,
> so it is arguable that persistence makes a timing error worse.
> You only have to watch a someone walk across the shot on a decent
> DVD player and then compare it will your typical soft decoder on
> a Windows laptop to know that seeing all the fields at the right
> time is a really important.
> 
> Interleaving only helps reduce apparent flicker. Each field represents
> a different point in time, so missing a field, or replacing one
> with a copy of the previous one is ruinous. Swap the fields and
> everything looks furry. 100Hz TVs don't just
> display each field twice; they have a DSP in there to
> sort out the motion compensation and synthesise the missing fields.
> My 100Hz TV still screws up on fast scrolling credits (horizontal
> and vertical).
> 
> Stick to what you know, Boyd.
> Except the guns; it's really, really tedious.
> 
> 
> 

Re: [9fans] A prick into the wasps' nest ;-)

[boyd, rounin]

> Stick to what you know, Boyd.

well that's what happens when you [i] try to drag up stuff i learnt 25 years ago.

RE: [9fans] A prick into the wasps' nest ;-)

[Andrew Simmons]

> 
> Even if I softly drop a few words here and there, this is still
> recognisably rubbish.
> 
> The frame rate is 25Hz for PAL, or 30Hz for NTSC (29.97 for pedants).
> SECAM is also 25 Hz.

Surely a shade harsh? And a pedant might claim that PAL, NTSC, and SECAM are
colour encoding systems, and as such independent of frame rate. Not to
mention the fact that Fowler prefers "recognize" to "recognise", and claims
that "synthetize", not "synthesize" is the correct formation.

> Stick to what you know, Boyd.

Oh right, so you have to know what you're talking about if you're going to
post to 9fans? If that were the case, then I'd certainly never ... ah, I see
what you mean. Better shut up now.

Re: [9fans] A prick into the wasps' nest ;-)

[Nigel Roles]

Andrew Simmons wrote:

>Surely a shade harsh? And a pedant might claim that PAL, NTSC, and SECAM are
>colour encoding systems, and as such independent of frame rate. Not to
>  
>
Yes, I think you'd have to be a pedant to claim that they weren't video 
standards.
Sorry if I appeared harsh; It was the preceding sequence of gun related
stuff that got me going; I was about to moan about it anyway, and then hit
this email.

>mention the fact that Fowler prefers "recognize" to "recognise", and claims
>that "synthetize", not "synthesize" is the correct formation.
>
>  
>
I have to admit that synthetize was a new one on me; I can now bore at 
parties
with that one. Looking back I see that 'interpolate' would have been a more
accurate word in the first place.

>Oh right, so you have to know what you're talking about if you're going to
>post to 9fans? If that were the case, then I'd certainly never ... ah, I see
>what you mean. Better shut up now.
>
>  
>
With a chap from Snell & Wilcox on the list I thought through what I 
posted, yes!