On Tue, Jan 31, 2006 at 04:58:25PM -0000, Oege de Moor wrote:
> what do you think?
Sounds great to me.
I also fully agree with what you wrote this morning (the need for
interesting benchmarks before we design fancy analyses).
Ondrej
> ---------------------------
>
> to pql group, ptql group, klaus havelund, darmstadt group, object/teams,
> douence-sudholt-fradet,
>
>
> Dear all,
>
> I'm writing to you because over the past few months we've
> been discussing instrumentation tools where the instrumentation
> is based on trace properties. There is a very large variety of
> such tools, all with different formalisms for specifying
> the trace properties, and with different implementation
> techniques.
>
> This stuff is very cool - but we need to assess its merits in
> practice. For example, one could argue that such instrumentation
> could be adequately done in existing AOP languages like Aspect/J
> and Caesar/J; inspecting the history is just not necessary and
> too costly at runtime... The only way to decide the question
> is a careful evaluation on a significant number of case studies.
>
> It would be a great help for all of us to have a central
> repository of such examples where trace-based instrumentation
> is applied to interesting systems. That repository can then be
> a start for studies of the expressiveness of the formalisms used
> to express trace properties, of the performance of different
> implementation techniques, and so on.
>
> Each example should come with:
> - a description of what it does
> - the specification for generating the instrumentation
> - the base program to be instrumented
> - scripts for doing the instrumentation
> - scripts for running the program
> - if your instrumentation tool is not publicly
> available on the web, an executable for it
>
> In first instance, the repository would be shared with
> all recipients of this message. The more different people
> contribute, the greater the value for us all, so do let me
> know if you feel I have missed someone out.
>
> If you are interested in collaborating on this repository,
> let me know, and we (the abc group) will set it up. I'm sure
> it will generate a lot of interesting work!
>
> Many thanks,
>
> -Oege de Moor
> (on behalf of the abc team)
>
>
>
>
>
>
>
Received on Tue Jan 31 17:55:52 2006
This archive was generated by hypermail 2.1.8 : Tue Jan 31 2006 - 18:30:10 GMT