RISCOS port of SDL runtime information
--------------------------------------

Last update 15th May 2003

Runtime requirements
--------------------

This library currently needs a minimum of RISCOS 3.6. The source code for the library (and a lot of the programs built with it) also need long file names.

To use the audio you also need 16 bit sound and to have installed the Digital Render module by Andreas Dehmel version 0.51 available from his web site:
  http://home.t-online.de/~zarquon

Note: As most programs ported from other OS's use high resolution graphics and a memory back buffer a machine with a StrongARM processor and 1 or 2MB of VRAM (or a better machine) is recomended.


RISCOS runtime parameters
-------------------------

Several environmental variables have been defined to make porting programs easier (i.e. By setting these variable you do not need to have source code
differences between OS's).

They are all defined on an application basis.

The <appname> used below is found as follows:
1. Use the name of the program unless it is !RunImage
2. Check the folder specification for the folder !RunImage is run from. If it is a folder name use that name, otherwise if it is an environmental variable of the form <XXX$Dir> use the value of XXX.

The variables are:

SDL$<appname>$TaskName - The name of the task for RISCOS. If omitted then <appname> is used for the task name,

SDL$<appname>$BackBuffer - set to 1 to use a system memory back buffer for the screen in full screen mode. Some programs on other systems assume their is always a back buffer even though the SDL specification specifies this is not the case. The current RISCOS implementation uses direct writes to the screen if a hardware fullscreen is requested.

SDL$<appname>$CloseAction - set the action for the close icon. Again as programs don't match the specification you can set this to 0 to remove the close icon from the main window for applications where this does not affect the program.

SDL$<appname>$AudioBuffer - set the minimum size of the audio buffer in samples in steps of 512. If this is less than the minimum calculated from the parameters requested in the program by SDL_OpenAudio these will be used instead. If this parameter is not used or set to 0 the buffer size will be calculated to buffer 10 centisecond of sound. The time buffered can be calculated as specified size/frequency in seconds. The larger the buffer the more audio data that will be buffered, but this can lead to a lag between a sound being requested and it being heard, so the size should be kept to a minimum. You should only set this parameter if the sound is being missed out. 
