Re: [avr-libc-dev] Automated testing project

[Joerg Wunsch]

As Peeter Vois wrote:

> I've been reading code of both approaches and would like to say my
> first blick opinion: Simulavr has much higher quality source code,
> it is understandable and well commented. Simulavrxxx is hard to
> understand and not well commented.

Yes, perhaps.  Simulavr is well structured, but often already ``too
well'', i.e. it became quite inefficient that way.  But I think the
main reason why Ted realized it was going to become a dead-end street
(besides he certainly already knew he had to give up some things due
to its health situation) is that it was close to impossible to add a
reasonable IO simulation using the approach taken, and/or to
improve the simulation speed.

In contrast, Klaus' simulavrxx was written to simulate a multi-CPU
model railway control, and he succeeded in that.  ISTR I once asked
Klaus about why he didn't extend simulavr, and he told me he tried but
had to give up because it turned out to become impossible to do.

Finally, simulavr has a total of 0 maintainers left, while simulavrxx
has at least one active developer plus a couple of additional
contributors (perhaps that's even an underestimation).

-- 
cheers, J"org               .-.-.   --... ...--   -.. .  DL8DTL

http://www.sax.de/~joerg/                        NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)