Re: [PATCHv2 1/2] perf tools: Collect tracing event data files directly

[Frederic Weisbecker <fweisbec@gmail.com>]

On Wed, Sep 28, 2011 at 10:03:08AM -0400, Steven Rostedt wrote:
> On Wed, 2011-09-28 at 15:55 +0200, Frederic Weisbecker wrote:
> > On Mon, Sep 26, 2011 at 04:56:06PM +0200, Jiri Olsa wrote:
> > > On Mon, Sep 26, 2011 at 09:36:31AM -0400, Steven Rostedt wrote:
> > > > On Mon, 2011-09-26 at 11:11 +0200, Jiri Olsa wrote:
> > > > > Changing the way the event files are searched by quering specified
> > > > > event files directly, instead of walking the events directory.
> > > > > 
> > > > > Hopefully this way is more straightforward and faster.
> > > > 
> > > > Have you looked at my code I posted earlier that uses the libparsevents?
> > > > 
> > > > It uses globs such that you could do -e sched:sched* and it will enable
> > > > all sched events.
> > > 
> > > ops, haven't seen those changes yet..
> > > I think I can go only with 2/2 patch, if the 1/2 collides with your changes
> > 
> > But it seems Steve's patches are not completely uncontroversial because
> > of some crazy disagreements on where the libparsevent.so should lay (tools generic
> > or tied to perf).
> 
> Which to me seems to be a silly road block, in which I never got a clear
> answer for.

Yeah we need to sort that out with Ingo.

> 
> > 
> > So until we get that situation solved, we should continue to move forward.
> 
> Sure, but it just forks the code even more, and it's almost to the point
> where maintaining a fork will just be easier, not to mention quicker to
> get out to the distros.

Right. May be Jiri can have a look at what trace-cmd does there.