Benjohn Barnes wrote: > > Having not even read the piece... > > I was following up on Software Transactional Memory from an earlier > Ruby Talk posting, and that looks extremely promising. Have you read this thread on Software Transactional Memory? http://patricklogan.blogspot.com/2007/02/misguided... It's about the downsides of STM. Regards, Michael
on 2007-02-12 12:27
on 2007-02-14 15:29
>> > > It's about the downsides of STM. Fantastic! :) I'd like something to temper my enthusiasm.
on 2007-02-14 16:56
> Have you read this thread on Software Transactional Memory? > > http://patricklogan.blogspot.com/2007/02/misguided... Well, I've had a browse, and it mostly seems to be a rant on the style of, 'I don't like it, it sounds dangerous, ooooo, it's horrible, look at it! These other people don't like it either!', without (that I've found yet) a clear statement of why. The main objection seems to be that software will end up being a mess of shared state that many process are franstically writing too, but that just seems like really bad design to me. Asynchronus message queues, it suggests, are a better option. I get the feel that there is a "holy war" between a functional approach, and an imperitive approach here, somewhere. I'll need to read it more closely, because that sounds interesting. I don't yet see why STM shouldn't be highly functional though (I personally like functional a lot). Cheers, Benjohn