[Haskell-cafe] Google Summer of Code - Lock-free data structures

[Florian Hartwig]

On 19 March 2012 00:59, Chris Smith <cdsmith at gmail.com> wrote:
> On Mar 18, 2012 6:39 PM, "Florian Hartwig" <florian.j.hartwig at gmail.com>
> wrote:
>> GSoC stretches over 13 weeks. I would estimate that implementing a data
>> structure, writing tests, benchmarks, documentation etc. should not take
>> more
>> than 3 weeks (it is supposed to be full-time work, after all), which means
>> that I could implement 4 of them in the time available and still have some
>> slack.
>
> Don't underestimate the time required for performance tuning, and be careful
> to leave yourself learning time, unless you have already extensively used
> ThreadScope, read GHC Core, and worked with low-level strictness, unpacking,
> possibly even rewrite rules.  I suspect that the measurable performance
> benefit from lockless data structures might be tricky to tease out of the
> noise created by unintentional strictness or unboxing issues.  And we'd be
> much happier with one or two really production quality implementations than
> even six or seven at a student project level.
>
> --
> Chris Smith

Thank you, Hofstadter's law definitely rears its head in many of my projects.
I do have some experience with ThreadScope and strictness issues, but
you I agree that I'm probably underestimating the time I need to
learn.
I also agree that my focus would be on quality rather than quantity. I
quite like the modularity of this project, because it minimises the
chance of having a lot of half-finished but useless code at the end of
summer.