Previewing RISC OS 4+ and Select - 4 - The Kernel :: My RISC OS

The Kernel has already been changed for the first Select CD but these changes will continue in future issues of the OS.

The Kernel will be leaner with components split from the Kernel. these (now externel) components will be easier to maintain as they can be modified and tested without a complete ROM build.

Components that have so far been extracted from the Kernel include:

Conversions (OS_Convert*SWIs)
SysVars (Sys$* variables)
OSSWIs (Miscellaneous high level SWIs)
- these were inappropriate for the Kernel in the first place
ReadLine (separate implementation of OS_ReadLine)

There are many minor improvements to the Kernel to improve functionality & stability.

Dynamic Areas can be treat as OS_Heap like areas.

Now applications can use a couple of OS_Dynamic Area calls instead of using RMA or writing a special memory manage for Dynamic areas.

Software writers can now give workspace pointers to system variables that are implemented by modules (code variables) and allowing them to use the variables reliably from C without large amounts of support code.

If shift is held down on Boot then a menu is produced that alows boot to be completed from a range of sources including:

harddrive
floppy drive
CD
network - various options - shared boots are shown
and the one that will give confidence to anyone running beta versions...
the boot can be returned to the original installed 4.02 / 4.03 / 4.04 boot in ROM.

As a non-programmer (my serious foray into program writing used punched tape as its main back-up device). I'd appreciate help in understanding the results of these changes.

Presumably the more that is removed from the Kernel, and written as a stand-alone component, the easier it will be to maintain, to find and remove redundant code and to write stable software for it.

Finally, again a question for software authors, what's the implication of all this for writing code that has to be used with OS3, 4.0 and 4+ systems?
ROL say that anything that will run on OS4 will run on OS4+ - will the opposite be the case? (surely not or OS4+ couldn't be easier for programmers).

Re: Previewing RISC OS 4+ and Select - 4 - The Kernel

(Score: 1)
by quatermass [5] (stuart@cybervillage.removethis.co.uk) on Aug 24, 2001 - 04:36 PM
(User information [6] | Send a message [7] http://acorn.cybervillage.co.uk/ [8])

In my opinion the reason for the stripping of the kernel has to be a result of Pace wanting a leaner OS.

By removing code which isn't required you get a smaller kernel which is easier to maintain. This is standard practise for small devices were only the absolute bare minimum of code is included.
This allows them to sell a device as cheaply as possible.
Perhaps Pace wish to see RISC OS used in an embedded device?

Re: Previewing RISC OS 4+ and Select - 4 - The Kernel

(Score: 1)
by runnyhunny on Aug 26, 2001 - 10:54 AM
(User information [9] | Send a message [10]

Do Pace have anything to do with this 'Select' stuff?

If a number of these improvements have come from Pace, then why don't we also see '32-bit compatible' in the feature list? Oh, and 'HAL for I/O controllers'?

Re: Previewing RISC OS 4+ and Select - 4 - The Kernel [11] by jc on Aug 28, 2001 - 12:51 AM
- Re: Previewing RISC OS 4+ and Select - 4 - The Kernel [12] by runnyhunny on Aug 28, 2001 - 11:55 AM

Operating System [1]: Previewing RISC OS 4+ and Select - 4 - The Kernel [2]

Re: Previewing RISC OS 4+ and Select - 4 - The Kernel

Re: Previewing RISC OS 4+ and Select - 4 - The Kernel