[m-dev.] STM rewrite

[Peter Ross]

On Wed, Nov 9, 2011 at 2:50 PM, Chris King <colanderman at gmail.com> wrote:
> Comments, suggestions, and questions welcome.  I hope eventually for
> this module to find its way into the standard library (no doubt
> pending stylistic changes).
>
It looks very interesting and the fact that it supports the non-C
grades is fantastic.

I'm curious that you have low level C code as well as a mercury
implementation, does this mean that STM performance is much better
under the C grades?

I much prefer retry_on_fail that peter wang suggested, may_block has
me thinking about blocking I/O so I really couldn't work out what it
was doing.

Also as a complete newbie to STM code it would be helpful to expand
the explanation of the following points, as they are not so clear to
me.

* retry() is not needed; Mercury already has a failure mechanism
* nested logs are not needed; Mercury already has backtrackable unique
modes (mostly_unique)
* or_else is not needed; Mercury already has if/then/else
* use of builtin failure allows determinism analysis to double as
wait-freedom check

Great job.

Pete

--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at csse.unimelb.edu.au
Administrative Queries: owner-mercury-developers at csse.unimelb.edu.au
Subscriptions:          mercury-developers-request at csse.unimelb.edu.au
--------------------------------------------------------------------------