[Virtualacorn-list] RISC OS 6 Installation Problem

[Jeremy Nicoll - virtualacorn]

Dave Symes <dave at triffid.co.uk> wrote:

> I'm now scrabbling in the dark, but I've done series of inserting Echo
> messages down the file, including one at the end and they all display
> before the freeze happens.

I run !Reporter on my system, starting very early in Boot, and one of the
reasons I have the Echo lines in files is that the echoed contents appear in
Reporter's log file as a constant reminder of what to expect.

If you were still trying to debug a freeze, one option with Reporter is for
it to write its log file out to disk a line at a time as it adds to it. Of
course this means anything logged that way is very slow, but it can be
useful. Of course having a low logging process running also changes the
timing interactions of everything else that's going on, and that could have
meant that a boot-time problem (when there's lots going on in parallel eg
hardware becoming ready for use as software tries to use it) mightn't
manifest itself the same way anyway.

On my RPCs I used to run only the barest possible boot process (as far as RO
was concerned) with only one app being 'seen'/'run' at the start - and that
was one of mine.  It then started all the other tasks, loaded modules etc
according to the order I wanted that done in, and did so with specific
delays between certain parts of the process.  It could be paused/ resumed/
stopped completely too.  It meant that the precise order in which things
happened was always the same, but could be changed if I wanted to see if
something else worked better...

In my VRPC system I've not yet implemented that as TBH I use VRPC less and
less (doing more under XP these days).  It's always annoyed me that when one
boots RO the monitor tells you very little about what's going in before the
desktop starts, and if there is an error (like the unimplemented SWI msg) it
can be hard to read it before the desktop starts.  Now I have quite a lot of
extra Echos in the process and a few places where I have to hit Enter to
have it continue, and I have changed the way some of the commands are issued
so that their normal output is not suppressed.  Now if my VRPC system fails
to boot I will know where it had got to.  


> I've also, in conjunction with the Echos, barred out each remaining line
> one at a time, rebooted... Frozen, un-barred, tried another... and so on
> In which case it seems, because of no change, and the final Echo displayed
> before the freeze, the problem is not there, but in whatever follows after
> !Boot.Resources.!Internet.!Run   For which I have no idea, as I don't know
> where the instructions that get run during the initial boot sequence are
> stored.

In !Boot, but quite a lot of it works by for example using the RO *Repeat
command to run a command against all the items in a directory or
sub-directory of boot.  This is much easier to see if you run Reporter and
then look at its log to see what commands are issued, when and why.



-- 
Jeremy C B Nicoll - my opinions are my own.