108612 - investigate the cost of sidebar

Status: RESOLVED FIXED

(SeaMonkey :: General, defect, P1)

Description

[Cathleen]

please provide a patch to remove sidebar to the bug for further analysis.

Comment 1

[Samir Gehani]

Posted some high-level measurements at:
<http://www.mozilla.org/xpapps/sidebar/perf/nav-effect.html>

Comment 2

[Samir Gehani]

Attachment: Patch to remove sidebar (on windows and unix).

Comment 3

[Blake Ross]

Technically <vbox id="sidebar-box" class="chromeclass-extrachrome"/> should 
also be removed.

Comment 4

[Blake Ross]

What does a patch like

Index: sidebarOverlay.js
===================================================================
RCS file: /cvsroot/mozilla/xpfe/components/sidebar/resources/sidebarOverlay.js,v
retrieving revision 1.84
diff -u -r1.84 sidebarOverlay.js
--- sidebarOverlay.js	2001/10/27 08:16:56	1.84
+++ sidebarOverlay.js	2001/11/06 04:13:24
@@ -517,6 +517,11 @@
 //////////////////////////////////////////////////////////////
 // Sidebar Init
 //////////////////////////////////////////////////////////////
+
+function sidebar_delayed_overlay_init() {
+  setTimeout("sidebar_overlay_init()", 0);
+}
+
 function sidebar_overlay_init() {
   if (sidebar_is_collapsed() && !gAboutToUncollapse)
     return;
@@ -1075,5 +1080,5 @@
 //////////////////////////////////////////////////////////////
 // Install the load/unload handlers
 //////////////////////////////////////////////////////////////
-addEventListener("load", sidebar_overlay_init, false);
+addEventListener("load", sidebar_delayed_overlay_init, false);
 addEventListener("unload", sidebar_overlay_destruct, false);

get us in terms of perceived speed? (Don't know if it would impact actual 
numbers.)

Comment 5

[Brendan Eich [:brendan]]

blake: you may know this, but it saves time spent running the JS compiler at the
call to setTimeout to use the function-calling alternative to the
string-eval'ing one that your example shows: setTimeout(sidebar_overlay_init,
0).  Optional args to the function come after the milliseconds (2nd) arg.  FYI,

/be

Comment 6

[Samir Gehani]

Cost has been investigated.  We are addressing this by implementing a tab
overflow solution so we paint less tab headers.  We found painting the tab
headers were taking a lot of time and affected the combination of new window +
page load time.