[lttng-dev] GSoC 2013, "C++ probe support" project and more

[Christian Babeux]

Hi Zifei,

> Besides the C++ support project, I'd like to work on the 'strace-alike
> wrapper'. It takes me a fair amount of time to try the demo of lttng.
> A 'strace-alike wrapper' is much more simpler to me. Can this be done
> by a simple shell/python wrapper of lttng-tools?

I do think so. Maybe David can chime-in here?

> "Improvement of liblttng-ust-libc-wrapper's coverage of libc's
> functions" looks interesting to me, however I need some time to be
> familiar with Dyninst.

The improvement of liblttng-ust-libc has nothing to do with dyninst.
Maybe you meant the dynamic instrumentation project?

> I've uploaded a patch to: http://bugs.lttng.org/issues/338

Could you post your patch to this mailing list following these guidelines:

- Indicate in the patch subject on which project the patch is to be
applied. In your particular case, the patches need to be applied on
the lttng-ust tree. You can accomplish this with the
'--subject-prefix' option when using the git format-patch command:

git format-patch -s -1 --subject-prefix="PATCH lttng-ust"

- Add your signoff in the patch. Use the '-s' option when using git
format-patch.

- Send the patch to the maintainer of the project and CC the list. In
your case, Mathieu Desnoyers (mathieu dot desnoyers at efficios dot com) is the
maintainer of lttng-ust. You can accomplish this with the following
git send-email command:

git send-email --to maintainer_email --cc list_email *.patch

I have a few minor comments on your patch, I will wait for the patch
to be posted on the mailing list so I can comment inline.

Thanks!

Christian

On Sat, Apr 20, 2013 at 8:29 AM, Zifei Tong <soariez at gmail.com> wrote:
>
> On Fri, Apr 19, 2013 at 1:38 PM, Zifei Tong <soariez at gmail.com> wrote:
> > While for g++, since it does not support C's designated initializer
> > (clang++ does), further work shall be done. All remaining errors are
> > related to this struct initializer issue
> > (https://gist.github.com/5kg/5417850), can anyone suggest a clean hack
> > to workaround this?
>
> I did some research on designated initializer today, and finally made
> g++ compile 'hello.cxx' example.
>
> G++ do support designated initializer, however only 'trivial
> designated initializers' are supported, otherwise it will complain:
> 'sorry, unimplemented: non-trivial designated initializers not
> supported'.
>
> After some trial-and-error, it seems that 'trivial designated
> initializers' means no out-of-order initialization and no missing
> initialization (except the fields on the tail of a struct).
> And nested initialization should be done in the form {.foo = {.bar =
> 1}} instead of {.foo.bar = 1}.
>
> So I reordered some initializers, add some fields, and change nested
> initializations in the above odd form to make g++ happy.
>
> I've uploaded a patch to: http://bugs.lttng.org/issues/338
>
> Could you please review it?
>
> Thanks.
> --
> Best Regards,
> 仝子飞 (Zifei Tong)
> College of Computer Science and Technology, Zhejiang University
>
> soariez at gmail.com / tongzifei at zju.edu.cn
>
> _______________________________________________
> lttng-dev mailing list
> lttng-dev at lists.lttng.org
> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev