Avoiding unrecoverable error in run time system

Once in a while applications runs into a “unrecoverable error in run time system” and gets killed.
Some causes triggering these problems I have seen are:
- heap corrupted
- attempting to free an already freed memory block
- out of memory while to allocate buffers for file

I have just got the last one today with the consequence that the file I was trying to update just got emptied as a result.
It happened on an old non-“RISC OS 5” machine I was playing with but chnces are that the same behaviour could happen on RISCOS 5 and got me writing this email.

I’m pretty sure that in most cases the Shared C library could avoid killing the application:
- heap corrupted, could be reported once in a error window, flagged as reported and then malloc() could return null, and free() return as if it worked.
- a similar approach could be used for attempting to incorrect calls free()
- file handling could revert to non-buffered mode when buffers cannot be allocated or only touch the file after it has successfully allocated its buffer.