To Code Slasher
----------------------
Thank you for testing... on real machines.
I hoped to have had fixed that damn mouse bug that show on your 200Mhz machine...
...obviousely i did not ...
Well, you could press Alt+3 to restart the PS/2 mouse driver -- as this is usually fixing it... i must be doing something "slightly" wrong there in the mouse initialization routine... since it is only showing on slower machines
Yes your testing did helped me.
Thanks again for testing.
To Tim Robinson
----------------------
Hehe so my "concepts" page did gotcha
Well i know your arguments, i did NOT posted that stuff on my site from ignorance.
I just happen to think othervise (when compared to you that is) and if this is of no interest to this forum i am prepared to take this debate on indeological terms via email
Email me if interested in this debate:
[email protected]
Maybe we can both understand our different oppinions better
Just a fast argumentation from my side:
------------------------------------------------
1)I do NOT see why a filesystem has to be unmounted... are you biased via unix style here? For example my OS beeing realtime and critical oriented i could just write thru all filesystem writes and use cache only when reading, or cache writes only for a very short period of time (for speedup) ..a time that any decent UPS system will keep power up (like a few secconds) -- of course offering an option to the user to choose this.
Also you did not understand that concept very well... or i did not expressed it right:
I did not mean "no runtime error checking" for results or something.. i still think that an algorithm/application should work well for all and every data thrown at it
,
What i mean was the useless checks for ring3 protection
What i am proposing is better checks at design time and lesser checks at run time...
(see point 3 below)
2) Keeping chache untrashed (and TLB/pagind) with prremptive multitasking is a hard job IMHO, also programming when you are interupted whenever OS feels like is not very efficient not very error prone IMHO
While most people praise preemptive multitaking for programmers ease of use -- i do not.
A matter of oppinion of course, -- as i well know the many advantages of preemptive multitasking ...
But my choice of oppinion is cooperative multitasking -- for personal reasons to long to express here....
3) I do NOT see why "ordinary" applications have to be "protectd" in what they can do.. i really see a use for this "sandbox" only when testing code... at developemnt time ... but at runtime this wastes preciouse resources with no real use... again a matter of oppinion that i am willing to sustain
Applications from "unprotected era" have/had far less bugs and crashes than today "protected" applications and environments
I do not intend to return to "old" times, insted i just want to design an OS that will be an example and sustain my concepts.
For making fair testing I have stated that I also intend to implement "standard" preemptive multitasking + paging + ring0/3 protection...
This way things could be fairly tested with same OS on both grounds
)