From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932744Ab3CLTwi (ORCPT ); Tue, 12 Mar 2013 15:52:38 -0400 Received: from mail-gh0-f174.google.com ([209.85.160.174]:46997 "EHLO mail-gh0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755090Ab3CLTwh (ORCPT ); Tue, 12 Mar 2013 15:52:37 -0400 Date: Tue, 12 Mar 2013 16:52:28 -0300 From: Arnaldo Carvalho de Melo To: Namhyung Kim Cc: Peter Zijlstra , Paul Mackerras , Ingo Molnar , LKML , Jiri Olsa , Andi Kleen , Namhyung Kim , Pekka Enberg Subject: Re: [PATCH 00/12] perf annotate: Add support for event group view (v2) Message-ID: <20130312195228.GC11779@ghostprotocols.net> References: <1362462812-30885-1-git-send-email-namhyung@kernel.org> <20130311205316.GA11779@ghostprotocols.net> <878v5tl2vc.fsf@sejong.aot.lge.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <878v5tl2vc.fsf@sejong.aot.lge.com> X-Url: http://acmel.wordpress.com User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Em Tue, Mar 12, 2013 at 12:21:27PM +0900, Namhyung Kim escreveu: > On Mon, 11 Mar 2013 17:53:16 -0300, Arnaldo Carvalho de Melo wrote: > > I need to go thru this more thoroughly, as at least one thing caught my > > attention, in some cases ->nr_pcnt is used and in others > > evsel->nr_members, do we need both? > Probably not. I was thought keeping an array and its length together is > better. But it's rather a waste to save the same information to every > date structure so I move it to the struct annotate_browser->nr_events. > For evsel->nr_members, current code sets it only for grouping is enabled > so the individual events will have value of 0. Also it's not pass the > evsel to every function need it. But if you think it should really be > fixed that way I can make the change to pass and use evsel->nr_members. This can be left for later, my point here is that we should try to avoid the various UIs from diverging too much, i.e. the code for --gtk, --stdio and --tui should be as close and using the same abstractions as possible. > Anyway, below is a fixed version that could survived my testing. indeed, works, compared the output of --stdio --group with --tui --group and it matches. While looking at it I noticed that the stdio output wastes the initial columns, at some point we may want to make the hist_entry lines match the TUI ones, so that going back and forth on two xterm tabs shows no differences. Anyway, thanks, applied! - Arnaldo