Re: [Chicken-users] types.db incorrect

[Jörg F . Wittenberger]

On Sep 8 2011, Felix wrote:

> > Unfortunately this situation paralyses me.  I too much depend on it.
>
> Does it work with specialization turned off? (-O2 only)

Dunno.  I went into punch-the-head-through-the-wall mode.
Got the problem fixed.
It turned out to be a abuse of code never intended to be used that way.
(I found that my llrb tree code stuff was not written to handle both
half-ordered key space and arbitrary deletes at the same time; but it
did not give a warning either when I began to use it that way.)
-- So much right now for the "unrelated" - but still related - bug
I've been writing about.

I'd like to take this time out to pre-announce an event, which took
place today around a quarter past two European time:

As far as my changelog goes back, there has been an serious ongoing
development effort to add chicken as an alternative to rscheme (in a way,
which boils down to some identical cryptographic check sums per operation)
since February 2009 (but the mail archive reveals that there have been a
seven more years of more cautious attempts) for http://ball.askemos.org/

Now, for the first time I've been able to replace the peers exposed to the
internet with the chicken based version.

I'm running now from a network like this:
3 nodes with FreeBSD and rscheme
two nodes Linux AMD64 and chicken
and one node ARM (Sheeva Plug) with Linux and chicken
plus a phantom node which does not exists (always fails).

The whole network is on average (computing their byzantine
agreement on transactions) 50% faster than before
(when only the ARM ran with chicken).

And it's running on approximately 50% of memory.

I guess it's time to prepare a first release eventually.

Thanks for all your precious help over all this time!

Have A Nice Day

/Jörg
.........