Re: [PATCH 5/5] perf buildid-list: New plumbing command

[Ingo Molnar <mingo@elte.hu>]

* Arnaldo Carvalho de Melo <acme@ghostprotocols.net> wrote:

> Em Mon, Nov 16, 2009 at 09:58:20PM +0100, Ingo Molnar escreveu:
> > 
> > * Arnaldo Carvalho de Melo <acme@ghostprotocols.net> wrote:
> > 
> > > Em Mon, Nov 16, 2009 at 07:58:52PM +0100, Frederic Weisbecker escreveu:
> > > > On Mon, Nov 16, 2009 at 04:32:45PM -0200, Arnaldo Carvalho de Melo wrote:
> > > > > With this we can list the buildids in a perf.data file so that we can
> > > > > pipe them to other, distro specific tools that from the buildids can
> > > > > figure out separate packages (foo-debuginfo) where we can find the
> > > > > matching symtabs so that perf report can do its job.
> > > > > E.g:
> > > > > [acme@doppio linux-2.6-tip]$ perf buildid-list | head -2
> > > > > 8e08b117e5458ad3f85da16d42d0fc5cd21c5869
> > > > > 520c2387a587cc5acfcf881e27dba1caaeab4b1f
> > > > > [acme@doppio linux-2.6-tip]$ perf buildid-list -v | head -2
> > > > > 8e08b117e5458ad3f85da16d42d0fc5cd21c5869 /sbin/init
> > > > > 520c2387a587cc5acfcf881e27dba1caaeab4b1f /lib64/ld-2.10.1.so
> > > > 
> > > > I would rather see the effect of this -v option as the default.
> > > > It's very useful to see the dso resolved but hashes alone
> > > > don't pay much.
> > > 
> > > I disagree, the main purpose of this command is equivalent to 'git
> > > rev-list':
> > > 
> > > [acme@doppio linux-2.6-tip]$ git rev-list HEAD^^..
> > > 089242a8c826aac1c827bb797475b9852fe0db49
> > > 18e3a75dc47d8397bbf0936ec8e3ebdecada7ba1
> > > [acme@doppio linux-2.6-tip]$ 
> > > 
> > > I.e. provide content keys that then will be used by distro specific
> > > porcelain to get the matching -debuginfo packages with the symtabs.
> > 
> > Hm, i'd say that's a (small) bug in Git.
> > 
> > Porcelain can add flags just fine to get 'raw' behavior - so in general 
> > we want to default the most human friendly parameters to the most 
> > human-readable output format.
> 
> Well, no strong feelings here, make it the other way around and the 
> most usual case will just ignore those extra characters :-)

i looked briefly but it didnt seem obvious to invert 'verbose' (which is 
library shared), and i have the excuse of it being late here - so i'll 
leave it up to you in a followup patch :-)

	Ingo