uvcceet schrieb:
> 
> --- In os2user{at}yahoogroups.com, Herbert Rosenau  wrote:
> 
> 
>>
>>SS HAS the right to change the underlying OS/2. It has done that by 
>>replacing the multimedia system, inserting ntfs.ifs in base install, 
>>extending the WPS in default, replacing Netscape with Mozilla,....
>>
>>There are lots of changes still outstanding. OS/2 gets renovated step 
>>by step.In eCS 1.0 it was the desktop, in 1.1 it was mainly only 
>>reinforcing something, in 1.2 a new multimedia subsystem to increase 
>>the usabiltiy. IBM itself will only make maintenance but brings 
>>nothing new.
>>
>>eCS 2.0 will come with the ability to boot from JFS - OS/2 can't.
>>
> 
> 
> Interesting. I was of the impression that eCS could not make kernel
> changes, but then I wouldn't know where the line is drawn between
> actual OS/2 code, and functional addons. I admit I don't really know
> any more than what has been bandied about. I don't view any of the
> changes eCS has made as being anything to the operatings system tho,
> as every one of them is just an add-on, not unlike what any user can do.
> 
> When do we get the SIQ problem fixed? :-)

Look on. IBM has delivered MCP2  more than 2 years ago. Since then 
there was FP4, peer fix, tcp/ip stack fix, tcp/ip fix, about 40 new 
os2ldr/os2krnl, USB fixes, LVM fix, JFS fixes, .......

eCS 1.2 english was released after IBM had consolidated all the 
kernlel/loader into a new stable state. The (originally faulty) FP4 
gots fixed because SS was unhappy with it.

German 1.2 gets delayed because translation was more costly than 
awaited - but in that time IBM was coming out with some base fixes 
like lvm, jfs, kernel, loader, ..... so the fixes gots included into 
the base 1.2 - extending the time needed to get transation done, 
system tested....

Now we have it. But chinese and other dbsc languages needs to get out....

No, it is impossible to make from the single system input queue a 
multy system input queue - it would break each and any PM program 
written since OS/2 1.1 beta.

The so called SIQ problem is NOT a SIQ problem but a problem of 
misdesigned programs. IBM had released a lot of guides how to hande 
the PM message queues (yes, plural) - but some developers were too 
dumb to understund the guides and so hungups occures.  As I had seen 
the original WPS 1.0 code it was more a quick hack than serious 
commercial programming. With WARP4.0 the WPS was written from scratch 
- having the problem to have to be compatible to the quick hack.

You are a developer? Write your PM applications strongly like the 
guides - and you will never see a problem with the SIQ - ignore the 
guides and you gets problems over problems.

The System Input Queue coordinates the data coming from 2 different 
input devices (keyboard and mouse) sending it seriell to the window 
havbing the focus at the time the data comes in. As each tread owns 
its own message queue the data gets moved from the System Input Queue 
into the queue of the thread that belongs to the window that had had 
the focus at the time the data was received from the device. But as 
many of the data (keypress, mouse button action,....) requires lots of 
actions in logical sequence - each message must be sended when the 
prior ones comes back. E.g.: key press -> to window. Window acts on 
and comes back with a result. Depending on the result there are other 
messages generated and sent out, coming back with a result, generating....

The corresponing message 'key up' has to wait until all actions based 
on 'key press' are done - else your program would act on 'key up' 
before 'key down' actions are finished - resulting in big chaos, eh?.

There is a rule that requires that no action on a message should 
lasted more than 1/10s - but some programmers are too dumb to follow 
that - result is: The system queue gets blocked, blocking anything.

In WARP4 FP5 the SIQ got fixed! Now each message seded from is sended 
with a time frame. When that ime runs out the application gets removed 
from the queue - so SIQ _is_ fixed. What you sees now is result from 
misdesigned hardware like IDE controller loosing interrupts, mainboard 
faults,........ resulting from the crappy hardware architecture of the PC.

I have here runnimng some computers unattached. One hungs itself every 
day multiple times - regardless which OS is active on it. I get the 
same with OS/2 WARP3, NT, M2K, WARP4, eCS 1,0, eCS 1.1).  You would 
say it is the SIQ problem - but how can Windows NT, Windows 2000, 
Linux, WARP3, WARP4, eCS all have the same problem - freesing the 
desktop while nobody uses mouse or keyboard but some applications 
running who hare update theyr windows periodically?

Based on that I see another computer running WARP4 unattached running 
for 3 das or 3 weeks - and then freesing the desktop. Another does the 
same with eCS 1.1 but only while up and running for 2 days to 3 month, 
another 2 weeks to 3 moth, another 2 days to 6 month while keyboard 
and mouse active 14h/day on it.

It is NOT the OS, it is the lousy hardware that hungs up the system 
periodically. Different chipsets on different mainboards with 
different graphic cards results in different uptime having the same 
OS, the same drivers, the same applications installed and active.

-- 
Tschau/Bye

Herbert Rosenau                    eCS 1.2 Deutsch bestellen bei
EDV Beratung & Programmierung           http://pc-rosenau.de
Eichelsdorfer Str. 15
97461 Hofheim i.UFr.

Tel: 49-9523-502834
Fax: 49-9523-502823                     http://www.dv-rosenau.de/


------------------------ Yahoo! Groups Sponsor --------------------~--> 
Make a clean sweep of pop-up ads. Yahoo! Companion Toolbar.
Now with Pop-Up Blocker. Get it for free!
http://us.click.yahoo.com/L5YrjA/eSIIAA/yQLSAA/E8folB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

 To visit your group on the web, go to:
    http://groups.yahoo.com/group/os2user/

 To unsubscribe from this group, send an email to:
    os2user-unsubscribe{at}yahoogroups.com

 Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 




---