Re: [RFC PATCH] perf tool: Add a global variable "const char *input_name"

[Feng Tang <feng.tang@intel.com>]

On Thu, 20 Sep 2012 12:05:51 -0300
Arnaldo Carvalho de Melo <acme@redhat.com> wrote:

> Em Thu, Sep 20, 2012 at 10:19:39PM +0800, Feng Tang escreveu:
> > Currently many perf ommands annotate/evlist/report/script/lock etc
> > all support "-i" option to chose a specific perf data, and all
> > of them create a local "input_name" to save the file name for
> > that perf data.
> > 
> > So adding a global variable could unify all those local ones, and
> > it will also be useful for other codes which need to know the file name.
> 
> But I think the input_name should be passed as a parameter to whoever
> needs it. Think about 'perf diff', for instance, it deals with two
> perf.data files.

Yes, when I wrote the patch, I checked all "perf_session__new()" calls, and
found "perf diff" uses the "input_new" and "input_old", so I thought it
should be immune to this change :)

The reason I came up with this patch is, when I rewrote the script browser
patch, I need to know this "input_name" to run script's report func on,
but IIUC currently this info is only saved in struct perf_session, and in
the browser stage (report/annotate), the perf_session is already phased
out and I don't know how to get a reference for the perf_session to get
the "input_name" 

> 
> In general I think we should _remove_ more globals :-)

Understood.

Thanks,
Feng