From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755050Ab2D3IcP (ORCPT ); Mon, 30 Apr 2012 04:32:15 -0400 Received: from LGEMRELSE7Q.lge.com ([156.147.1.151]:61615 "EHLO LGEMRELSE7Q.lge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753255Ab2D3IcN (ORCPT ); Mon, 30 Apr 2012 04:32:13 -0400 X-AuditID: 9c930197-b7ca1ae0000069e6-56-4f9e4e073af6 From: Namhyung Kim To: Pekka Enberg Cc: Arnaldo Carvalho de Melo , Peter Zijlstra , Paul Mackerras , Ingo Molnar , LKML Subject: Re: [PATCH 7/7] perf ui/gtk: Use struct perf_error_ops References: <1335761711-31403-1-git-send-email-namhyung.kim@lge.com> <1335761711-31403-8-git-send-email-namhyung.kim@lge.com> Date: Mon, 30 Apr 2012 17:31:07 +0900 In-Reply-To: (Pekka Enberg's message of "Mon, 30 Apr 2012 09:31:28 +0300 (EEST)") Message-ID: <87ipghsjsk.fsf@sejong.aot.lge.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.95 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-Brightmail-Tracker: AAAAAA== Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi, On Mon, 30 Apr 2012 09:31:28 +0300 (EEST), Pekka Enberg wrote: > On Mon, 30 Apr 2012, Namhyung Kim wrote: >> Define and use perf_gtk_eops to provide a GTK2 message >> dialog for error reporting. To do that, we need global >> main_window variable for tracking UI state. >> >> Signed-off-by: Namhyung Kim >> diff --git a/tools/perf/ui/gtk/util.c b/tools/perf/ui/gtk/util.c >> index a727fe394e91..0d0ed2ed3937 100644 >> --- a/tools/perf/ui/gtk/util.c >> +++ b/tools/perf/ui/gtk/util.c >> @@ -2,6 +2,53 @@ >> #include "../../util/debug.h" >> #include "gtk.h" >> >> +#include >> +#include >> +#include >> + >> + >> +GtkWidget *main_window; >> + >> +static int message_dialog(const char *title, const char *format, va_list args) >> +{ >> + char *msg; >> + GtkWidget *dialog; >> + GtkMessageType message_type = GTK_MESSAGE_WARNING; >> + >> + if (!main_window || vasprintf(&msg, format, args) < 0) { >> + fprintf(stderr, "%s:\n", title); >> + vfprintf(stderr, format, args); >> + return -1; >> + } >> + >> + if (strcmp(title, "Error") == 0) >> + message_type = GTK_MESSAGE_ERROR; >> + >> + dialog = gtk_message_dialog_new_with_markup(GTK_WINDOW(main_window), >> + GTK_DIALOG_DESTROY_WITH_PARENT, >> + message_type, >> + GTK_BUTTONS_CLOSE, >> + "%s\n\n%s", title, msg); >> + gtk_dialog_run(GTK_DIALOG(dialog)); >> + gtk_widget_destroy(dialog); >> + free(msg); >> + return 0; >> +} > > I think this is an usability glitch waiting to happen - especially so if > you use it for warnings. There's no reason to require the user to react to > warning messages in the GUI because there's absolutely nothing they can do > about them. > > I guess we could do something like the "ui helpline" thing used by the > newt front-end if we really wanted to. > > Pekka I did quick grep on ui__warning and found that most of its users are trying to show the messages before exiting. I think some (at least) of them can be converted to ui__error(). And as existing implementation (TUI) already requires user input for this, I thought it's ok. But I agreed with you that ui__warning should not be used for showing non-critical messages and converted to helpline-style ones. Thanks, Namhyung