[MERGE][bug #102019] strace hanging the test suite

[Martin Pool]

On 7/3/07, Vincent Ladeuil <v.ladeuil+lp at free.fr> wrote:

>     >> def strace_detailed(function, args, kwargs, follow_children=True):
>     >> ...
>     >>
>     >> and strace becomes
>     >> def strace(function, *args, **kwargs):
>     >> return strace_detailed(function, args, kwargs)
>
> Thanks for the hint, I didn't know how to do it in python.
>
>     >>
>     >> and the tests that test executing strace can do
>     >> strace_detailed calls rather than strace calls.
>
>     mbp> I agree with Robert that this style is nicer.
>
> strace_detailed looks ugly to my eyes, but here is a patch that
> implements your suggestion.

It's more explicit - it keeps separate that "this is passed to the
traced function" and "this controls the tracing".  I agree the
declaration is uglier.

>
>     mbp> I know there are cases in the test suite where we mix
>     mbp> them together and that's why I have this opinion.
>
> I can't find them or I do not understand what you are talking about ;)

I mean, that run_bzr has kwargs some of which control the running and
some of which are passed through.  Or at least it used to.

>
>     mbp> However, it seems to me that using strace -f on bzr is
>     mbp> likely to cause hangs in regular operation just as in
>     mbp> the test suite.  Therefore I don't see it as urgent to
>     mbp> support it until it actually works properly.

Yes, you're right, I'd forgotten it was only the presence of threads
that provokes the problem.

-- 
Martin