Greetings.

I have used different settings and stuff on previous versions, with what I thought were noticable improvements, although that is hard to prove really.

I am wondering if there are any speed settings for v1.5 that really work. I have a pretty fast machine, e6750,4g ram, raid 0, and a 3mb x 512 dsl connection. Pages load pretty fast. I have tried all of the settings in different combo's in this thread

http://kmeleonbrowser.org/forum/read.php?1,10095

I am wondering
A) if these still apply to v1.5 without any changes.
if the file user.js is where to apply them
C) if anyone has setting they are using in v1.5 that they think they see improvement with

I am testing, if this is considered the way to do it, by disabling all caching options. Adding entries to user.js, starting KM, then going to websites that I notice a considerable amount of time to get up and displayed, the longest IMO is www.tomshardware.com/us
this site even on this computer, using KM,Opera,FF2,FF3 or IE is fairly slow to load up. Compared to other sites that display pretty fast. I needed a site somewhat slow to see if an appreciable difference can be seen.

The most improvement I have seen is when I use the Proxomitron. Of course getting rid of the ad images can improve any sites load time because of not having to fetch off-site resources.

Does anyone have any information to share regarding this?

Many thanks in advance.

MrWoo

I am using the Speedup macro from the MacroLibrary. I have not checked all the items in the thread you noted against it. Sounds like there are many things to check/change and kind of hard to get an exact time for page load speed comparison.

This is exactly where I am at. I have tried many combinations, and some allow the page to say pause a few seconds, gather all the resources, and then just pop it all up at once, which 'looks' faster. Yet with no tweaks, the page starts loading parts, and while it 'looks' like it takes longer, the actual time seems the same.

lol, trying to combine those in that thread to the right mix is taking awhile.

MrWoo.

Is there some way to know that the method of loading is the same for each try? What I mean is if one "test" loads from the internet and another "test" loads from cache, they will be different. Are you trying to include the effects of your changes on your internet connection in the tests?

I have started work on something to quantify the load time, but then I thought of cache and I was not sure how to proceed.

I was going to start a new topic, but I will just put my question here.

Does KM read the cofiguration file only once at startup? Specifically, I need to change the cache use value only for the duration of a macro. I would like to force the page to load without using cache, but I don't want to destroy the user's current cache.

Browser.cache.check_doc_frequency is default at integer 3. Integer 1 would be Check for a new version every time a page is loaded. I am not sure if that means it would reload the page from the net for sure.

These are very good thoughts, the same I have.

So I am looking for some sort of quantative difference. I can create scripts to watch resource usage for the process, or watch network activity. Or I can have a button stop the timer when I feel the webpage is loaded. I am unsure which would be the best way, and even more so, I am unsure the result would be great enough to tell. Face it when the timer returns a 1.17 sec difference, I am not going to realize that.

So to me, testing required cache being off completely, and all cache cleared, to nothing. This now forces the fetching of all files needed to render a page.

I have to be able to see and feel the difference for it to be a real difference. IMO what seems like 3 seconds is probably more than that. Or more than likely, what will seem like instantaneous. A second delay and then the entire page rendered instead of it filling in over a number of seconds.

As to whetehr or not my connection plays a role, it is unknown to me how to capture this data. Once can use maybe process explorer, and watch for spikes in the graphs and data.

I am not ambitious enough yet to script up a parser for tdimon logs. This will tell the actual times of net activity, so your conclusions can be based more off fact.

I don't explicitly understand the methods used to input parameters into preferences though. At work I tried to open prefs.js and input some values. Then save prefs.js and then start KM. (km process is completely killed between different settings input). This did not work, as always the prefs.js was replaced. At home then I tried using user.js. This allowed the prefs to go in ok. However, after playing much, KM started to become very sluggish. Inspection showed that every value used in user.js was imported into prefs.js. This makes for some wierd performance lol!

So, I did a comparison with Opera loading tomshardware. Opera has always been an impressive renderer. I am not pleased with the direction it is heading, too much like all other 'browser suites'. So I focus more on KM, as it is still what Opera used to be. Insanely fast and not trying to assimilate the universe. For the eye to see, Opera is still just a very small amount noticable faster than KM.HEADS UP DEVS !!! This is a VERY VERY good thing. I offer congratulations to even be so good as to seem so close. Of course, this is only my opinion.

So, anything to make it appreciably faster is only icing on the already delicious cake.

MrWoo.

Quote
MrWoo
speed settings for v1.5 that really work.

Keep in mind that I am on dialup. But I have tried many tweaks (and combinations) at http://www.extremetech.com/article2/0,2845,1854508,00.asp
as well as others including the good old standard page speed at the macro library. But eventually I often end up getting "Page can't be found."
I aways end up going with out any tweaks, and all is fine. In fact, I don't think any of it was of any help any way. It's as if 1.5's speed just cannot be improved upon. Besides, on dialup, I thinks it may be the fastest browser alive just as it is.

Charlie

~~If it ain't broke, why screw it up?~~

I have created two small macro and js files to report the my load time. To use this you need to make the assumption that the time it takes to exec the js file is the same for each test. The macro executes the js to obtain a time, opens the URL, and based upon the on-load event, executes the js again for the time. It then reports the difference in the time.

I have from time to time when using PgLdSound.kmm found pages that tripped the on-load event more than once. I ran three pages twice to see the effect of having them in cache. This is my results. I am on a DSL line and live close to the craigslist site. These pages were not in my cache on the first try. I used my 1.1.6 version KM to obtain the URLs.

I did a preview of this post and I see that the wiki/forum software is not resolving the codes for bold and code.

http://charlotte.craigslist.org/tls/  0.89  0.547
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx  3.453  2.109
http://www.alltheweb.com/    4.063  0.922

LoadTime.kmm

#  K-Meleon Macros (http://kmeleon.sourceforge.net/wiki/index.php?id=MacroLanguage)
#
# ---------- LoadTime.kmm
# ---------- K-Meleon Macro Language to poll the time before and after pageload ---------------
#
# Dependencies		: main.kmm
# Resources				: LoadTime.js
# Preferences			: -
# Version					: 0.5   2008-10-05
# --------------------------------------------------------------------------------

LoadTime_RunCode {
	setpref( BOOL, "k-meleon_loadtime_active", TRUE);
	$_LoadTime_URL = prompt("Enter or paste here the URL for the page to test.","Website URL");
	exec("wscript.exe \"".$_LoadTime_Path."\" 1");
	open( $_LoadTime_URL ); 
	}
	
_LoadTime_Loaded_Run {
	getpref( BOOL, "k-meleon_loadtime_active" )== true ? &_LoadTime_Loaded_RunTrue : 0;
	}

_LoadTime_Loaded_RunTrue {
	exec("wscript.exe \"".$_LoadTime_Path."\" 2") ;
	togglepref( BOOL, "k-meleon_loadtime_active") ;
	}	
_LoadTime_BuildMenu{
	setmenu("&Tools",macro,"LoadTime Test",LoadTime_RunCode,Misc);
	}

_LoadTime_GetPath{
### LoadTime.js is supposed to be located in the same folder as LoadTime.kmm (this file).
### If it isn't, Windows Script Host will notify the user.
### Double underscore makes variable local only.
$__Data=readfile(getfolder(MacroFolder)."\\LoadTime.kmm");
$_LoadTime_Path=getfolder($__Data==""?UserMacroFolder:MacroFolder)."\\LoadTime.js";
}

$OnLoad=$OnLoad."_LoadTime_Loaded_Run;";
$OnStartup=$OnStartup."_LoadTime_GetPath;";
$OnInit=$OnInit."_LoadTime_BuildMenu;";
$macroModules=$macroModules."LoadTime;";

LoadTimejs

/* LoadTimejs by JamesD, version  0.5   2008-10-05

   This script is a helper application for the K-Meleon Macro Module LoadTime.kmm that
   produces an elasped time for a page load.
   
   Windows Script Host 2.0 (or better) required!
   
   WSH is part of Microsoft Internet Explorer 5.0 and newer. Latest version of WSH is
   available at http://msdn.microsoft.com/scripting/
*/
var d, t ;
if(WScript.Arguments.count()==1)
{
	var WshShell = new ActiveXObject("WScript.Shell")
	if (WScript.Arguments(0) == 1)
		{
			d = new Date();
			t = d.getTime();
			WshShell.RegWrite ("HKCU\\Software\\KMELEON\\LoadTime\\start", t, "REG_SZ");
		}
	if (WScript.Arguments(0) == 2)
		{
			d = new Date();
			t = d.getTime();
			WshShell.RegWrite ("HKCU\\Software\\KMELEON\\LoadTime\\finish", t, "REG_SZ");
			var $_LoadTime_Start = WshShell.RegRead ("HKCU\\Software\\KMELEON\\LoadTime\\start") ;
			var $_LoadTime_Finish = WshShell.RegRead ("HKCU\\Software\\KMELEON\\LoadTime\\finish") ;
			var $_LoadTime_Elapsed = $_LoadTime_Finish - $_LoadTime_Start ;
			var $_LoadTime_Seconds = $_LoadTime_Elapsed / 1000 ;
			WScript.Echo( $_LoadTime_Seconds + " Seconds elapsed time") ; 
			

			WshShell.RegDelete ("HKCU\\Software\\KMELEON\\LoadTime\\start") ;
			WshShell.RegDelete ("HKCU\\Software\\KMELEON\\LoadTime\\finish") ;
		}		
}
else
{
objArgs = WScript.Arguments ;
WScript.Echo(WScript.Arguments.Count());
for (i=0; i<objArgs.length; i++)
	{
    	WScript.Echo(objArgs(i))
	}
}

Edited 1 time(s). Last edit at 10/05/2008 01:29PM by JamesD.

Very nice. I will try this out tommorrow and report my findings.

MrWoo.

All Gecko's use more or less the same speed twweaks. Here is an old list.
AFAIK It is also in Forum history.

Excerpts from Firefox tweak guide created by Achilles

Most of us have heard of these tweaks, and they can be used with any Mozilla.
The settings are inserted into user.js, which feeds them into prefs.js on start.
Achilles based his summary at Tweakfactor on a thread at Mozilla Forum.

K-Meleon and some other Gecko browsers that use their native OS widgets will not use tweaks for XUL items.
- So not all tweaks are applicable for k-m, but even those that are not usefull do no harm; k-m just overlooks them.
Since K-Meleon goes with low resources - Slow Computer means slower than Pentium 500 for us ;-) .
Quick and Dirty Settings

user_pref("network.http.pipelining", true);
user_pref("network.http.pipelining.firstrequest", true);
user_pref("network.http.pipelining.maxrequests", 8);
user_pref("nglayout.initialpaint.delay", 0);

Pipelining does multiple data requests at once and should speed things up. Initial Paint Delay actually slows down the rendering of the ENTIRE page but I tend to start reading before the entire page is rendered ; - ) this gives me the impression that the page loads faster.
Common to all configurations

These are the settings that seem to be common to all configuration files regardless of connection speed or computer speed with a couple of additions - plugin paths can be found with about:plugins and the bookmark menu delay is turned off.

user_pref("network.http.pipelining", true);
user_pref("network.http.proxy.pipelining", true);
user_pref("network.http.pipelining.maxrequests", 8);
user_pref("content.notify.backoffcount", 5);
user_pref("plugin.expose_full_path", true);
user_pref("ui.submenuDelay", 0);

Fast Computer Fast Connection

user_pref("content.interrupt.parsing", true);
user_pref("content.max.tokenizing.time", 2250000);
user_pref("content.notify.interval", 750000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 750000);
user_pref("nglayout.initialpaint.delay", 0);
user_pref("network.http.max-connections", 48);
user_pref("network.http.max-connections-per-server", 16);
user_pref("network.http.max-persistent-connections-per-proxy", 16);
user_pref("network.http.max-persistent-connections-per-server", 8);
user_pref("browser.cache.memory.capacity", 65536);

A couple settings of note - Firefox is allocated 4096 KB of memory by default and in this configuration we give it roughly 65MB as denoted by the last line. This can be changed according to what is used.

Fast Computer, Slower Connection

This configuration is more suited to people without ultra fast connections. We are not talking about dial up connections but slower DSL / Cable connections.

user_pref("content.max.tokenizing.time", 2250000);
user_pref("content.notify.interval", 750000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 750000);
user_pref("network.http.max-connections", 48);
user_pref("network.http.max-connections-per-server", 16);
user_pref("network.http.max-persistent-connections-per-proxy", 16);
user_pref("network.http.max-persistent-connections-per-server", 8);
user_pref("nglayout.initialpaint.delay", 0);
user_pref("browser.cache.memory.capacity", 65536);

Fast Computer, Slow Connection(dial-up)
user_pref("browser.xul.error_pages.enabled", true);
user_pref("content.interrupt.parsing", true);
user_pref("content.max.tokenizing.time", 3000000);
user_pref("content.maxtextrun", 8191);
user_pref("content.notify.interval", 750000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 750000);
user_pref("network.http.max-connections", 32);
user_pref("network.http.max-connections-per-server", 8);
user_pref("network.http.max-persistent-connections-per-proxy", 8);
user_pref("network.http.max-persistent-connections-per-server", 4);
user_pref("nglayout.initialpaint.delay", 0);
user_pref("browser.cache.memory.capacity", 65536);

Slow Computer, Fast Connection
user_pref("content.max.tokenizing.time", 3000000);
user_pref("content.notify.backoffcount", 5);
user_pref("content.notify.interval", 1000000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 1000000);
user_pref("content.maxtextrun", 4095);
user_pref("nglayout.initialpaint.delay", 1000);
user_pref("network.http.max-connections", 48);
user_pref("network.http.max-connections-per-server", 16);
user_pref("network.http.max-persistent-connections-per-proxy", 16);
user_pref("network.http.max-persistent-connections-per-server", 8);
user_pref("dom.disable_window_status_change", true);

One of the changes made for this particular configuration is the final line where the status bar is disabled for changing web pages to save processor time.

Slow Computer, Slow Connection(dial-up)

user_pref("content.max.tokenizing.time", 2250000);
user_pref("content.notify.interval", 750000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 750000);
user_pref("nglayout.initialpaint.delay", 750);
user_pref("network.http.max-connections", 32);
user_pref("network.http.max-connections-per-server", 8);
user_pref("network.http.max-persistent-connections-per-proxy", 8);
user_pref("network.http.max-persistent-connections-per-server", 4);
user_pref("dom.disable_window_status_change", true);

That macro with js file is a really nice tool for this stuff. I will test a bunch tonight and tommorrow and post the results. Is there a way to just make it active on a typed in URL instead of having to use the menu option and pasting in what you want?

MrWoo.

@ Mr Woo

I wanted as little macro processing as possible for testing load time. That is why I had the URL entered via a prompt. It might be possible to read the $URLBAR variable instead of the prompt. I will have a look as soon as I get time.

@ Mr Woo

Try this change. Replace the commented line below with the one above it.

	$_LoadTime_URL = $URLBAR ;
	#$_LoadTime_URL = prompt("Enter or paste here the URL for the page to test.","Website URL");

Here are my tests so far. Thanks JamesD for that URL fix. Made it a lot easier. I was going to try and place it as a button, but frankly don't have time right now to learn the syntax or for that matter make an icon.

Tests went like this. Cleared cache, disabled all caching. Close KM. Open KM. Type in URL exactly as in the tests, used LoadTimeTest, record. Used LoadTimeTest again (4 times in a row). Record. Clear cache, enable default cache settings. Close KM. Open KM. Repeat. Same sequence for the tweaks, added them. Then removed them.

Here are my results so far. Pretty inconclusive for the most part. It really depends on the content of the site I guess as to wehter the tweaks or proxomitron do much.




--------------------------------------------------------------------------------------
default settings: no cache of any kind, flashblock enabled
http://tomshardware.com/us 11.62 8.94 9.07 8.79
http://anandtech.com 12.19 11.53 11.39 11.37
http://newegg.com 4.82 4.46 4.51 4.46
http://theregister.co.uk 7.73 6.67 6.92 6.67
http://majorgeeks.com 5.54 5.28 5.23 5.28
http://charlotte.craigslist.org/tls .74 .65 .65 .65
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 5.64 5.40 5.21 5.40
http://alltheweb.com .90 .82 .82 .84

default settings, default cache, flashblock enabled
first number initial load, second with cached content
http://tomshardware.com/us 10.43 6.23
http://anandtech.com 10.21 3.79
http://newegg.com 3.79 1.57
http://theregister.co.uk 8.09 2.87
http://majorgeeks.com 2.12 1.29
http://charlotte.craigslist.org/tls .64 .10
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 1.98 .90
http://alltheweb.com 3.26 .32
--------------------------------------------------------------------------------------------------
default settings, no cache, flashblock enabled, using Proxomitron, no tweaks
http://tomshardware.com/us 9.15 8.79 8.89 8.96
http://anandtech.com 8.53 7.79 7.71 7.79
http://newegg.com 6.68 3.53 3.32 3.78
http://theregister.co.uk 3.57 3.29 3.35 3.29
http://majorgeeks.com 3.39 3.18 3.11 3.12
http://charlotte.craigslist.org/tls .65 .61 .65 .65
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.15 3.34 3.36 3.56
http://alltheweb.com .76 .71 .62 .67

default settings, default cache, flashblock enabled, using Proxomitron, no tweaks
http://tomshardware.com/us 12.03 2.46
http://anandtech.com 5.45 2.38
http://newegg.com 6.31 1.21
http://theregister.co.uk 3.65 1.12
http://majorgeeks.com 2.14 1.29
http://charlotte.craigslist.org/tls .67 .56
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.36 1.5
http://alltheweb.com 3.73 .31
---------------------------------------------------------------------------------------------------
default settings, no cache enabled, flashblock enabled
Using this set of tweaks
*******
user_pref("content.interrupt.parsing", true);
user_pref("content.max.tokenizing.time", 2250000);
user_pref("content.notify.interval", 750000);
user_pref("content.notify.ontimer", true);
user_pref("content.switch.threshold", 750000);
user_pref("nglayout.initialpaint.delay", 0);
user_pref("network.http.max-connections", 48);
user_pref("network.http.max-connections-per-server", 16);
user_pref("network.http.max-persistent-connections-per-proxy", 16);
user_pref("network.http.max-persistent-connections-per-server", 8);
user_pref("browser.cache.memory.capacity", 65536);
********
http://tomshardware.com/us 9.93 9.64 10.21 9.95
http://anandtech.com 6.5 6.14 6.21 6.14
http://newegg.com 3.43 3.32 3.18 3.37
http://theregister.co.uk 7.04 6.56 7.01 6.76
http://majorgeeks.com 4.20 2.59 2.59 2.61
http://charlotte.craigslist.org/tls .65 .70 .65 .65
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.92 3.56 3.54 3.68
http://alltheweb.com .90 .85 .84 .62

default settings, cache at defaults, flashblock enabled
http://tomshardware.com/us 9.59 6.04
http://anandtech.com 6.57 3.65
http://newegg.com 3.28 1.54
http://theregister.co.uk 5.23 2.86
http://majorgeeks.com 2.43 1.29
http://charlotte.craigslist.org/tls .64 .06
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.82 .85
http://alltheweb.com .86 .31
---------------------------------------------------------------------------------
default settings, no cache, flashblock enabled
using these tweaks
*****
user_pref("content.interrupt.parsing", true);
user_pref("content.max.tokenizing.time", 30);
user_pref("content.maxtextrun", 8190);
user_pref("content.notify.backoffcount", 20);
user_pref("content.notify.interval", 10);
user_pref("content.notify.ontimer", true);
user_pref("content.notify.threshold", 100);
*****
http://tomshardware.com/us 9.92 9.17 9.29 12.01
http://anandtech.com 11.85 11.23 11.53 11.20
http://newegg.com 4.78 4.48 4.51 4.51
http://theregister.co.uk 7.33 6.67 7.03 6.79
http://majorgeeks.com 5.46 5.00 5.31 5.14
http://charlotte.craigslist.org/tls 1.07 .75 .76 .75
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 5.51 5.31 5.18 5.26
http://alltheweb.com .82 .70 .93 .70

default settings, cache at defaults, flashblock enabled
http://tomshardware.com/us 9.56 5.78
http://anandtech.com 7.06 3.75
http://newegg.com 4.45 1.60
http://theregister.co.uk 5.45 2.96
http://majorgeeks.com 3.96 1.07
http://charlotte.craigslist.org/tls .78 .04
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.70 .92
http://alltheweb.com .93 .32
--------------------------------------------------------------------------------
default settings, no cache, flashblock enabled, using Proxomitron
using these tweaks
****
user_pref("network.http.pipelining", true);
user_pref("network.http.pipelining.firstrequest", true);
user_pref("network.http.pipelining.maxrequests", 8);
user_pref("nglayout.initialpaint.delay", 0);
****
http://tomshardware.com/us 9.21 8.62 8.79 8.70
http://anandtech.com 8.59 7.71 7.81 7.71
http://newegg.com 3.64 3.57 3.43 3.32
http://theregister.co.uk 3.68 3.32 3.31 3.31
http://majorgeeks.com 3.51 3.00 3.28 3.01
http://charlotte.craigslist.org/tls .68 .64 .65 .65
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.45 3.43 3.28 3.48
http://alltheweb.com .73 .61 .60 .59

default settings, default cache, flashblock enabled, using proxomtron
http://tomshardware.com/us 11.19 2.26
http://anandtech.com 5.29 2.36
http://newegg.com 3.53 1.15
http://theregister.co.uk 3.51 1.12
http://majorgeeks.com 2.11 1.14
http://charlotte.craigslist.org/tls .73 .53
http://msdn.microsoft.com/en-us/library/72bd815a(vs.80).aspx 3.43 1.04
http://alltheweb.com .75 .31
------------------------------------------------------------------------------------

@ MrWoo

Quote

Pretty inconclusive for the most part.

Wow, you sure did a lot of testing. I am glad the fix for the macro helped. I will look into the button idea.

The inconclusive part may cause me to drop my next project. That was to display about 48 prefs with the default setting, the user's current setting, some recommended settings based on computer/connection speed, a link to the Mozilla site that explains the setting, and a comment/warning field.

Now I am not sure that it would be worth my effort.

@ MrWoo

I think I have the button worked out. I have put some clock buttons into a 7z file on my website.

http://www.perigee.net/~jmd8421r/Clock_bmp.7z

This is the code that I added to my toolbars.cfg file. I am using the Klassic skin and I put this code right after the code for Print.

LoadTime{
macros(LoadTime_RunCode)
Run LoadTime macro on current $URLBAR
hourglass.bmp
hourglass.bmp
hourglass.bmp
}

Ok. I will try that. Maybe here is an idea as well. As times are a little off each time, can this be automated, where I can put in a list of URL's, and then have the script load each x number of times. I know just enough vbs to build a fso and pipe out the results, but it would appear that you know it better than I and might be able to add that?

So that in script, var1 = url1, var2 = url2 etc, Then a For x = 1 to however many URL vars there are.

Then an output to text file of the URL and each time returned. The cscript msgbox is ok to keep, as it allows you to for instance pause between loadtimes. Or, a sleep function could be inserted for a few thousand milli's.

This would allow faster testing of things, and facilitate more tweaks to be used to see a broader range of differences. I only used I think 2 or 3 sets, but there are many possible combos. It takes awhile one at a time though.

MrWoo

It sound more easy than it would actually be to do that. We don't have a sleep/pause on the macro side and the ability for WScript to communicate back to KM is not easy at all. See this thread
http://kmeleonbrowser.org/forum/read.php?4,84281

Situation is this. The macro calls WScript to get the time and store it in a registry value. The macro then starts the loading process. When the load is finished, the on-load event has the macro again call WScript to get the time again. When WScript has both times, it does a substraction and presents the difference. There is no good way yet to tell the macro to wait until WScript is finished before starting another load.

I have taken a different approach, sort of a workaround. I have not looked yet at how to set up a list of sites to run through programatically. But I do have your button working, thanks! And I dug out the vbs.chm file and came up with a method to write to a fso textstream (aka write to logfile lol). It seems to work so far. I also set the statusbar to notify when the test is complete. Next I think I will figure out how to either open the logfile so you can create a divider and show your tweaks, or maybe some kind of prompt. Been a long time since I messed with vbs, and I have never messed with Jscript, so it was a learning experience.

Thanks JamesD for getting it rolling. Here is what I have now, including your button working.

THE MACRO::: LoadTime.kmm

# K-Meleon Macros (http://kmeleon.sourceforge.net/wiki/index.php?id=MacroLanguage)
#
# ---------- LoadTime.kmm
# ---------- K-Meleon Macro Language to poll the time before and after pageload ---------------
#
# Dependencies : main.kmm
# Resources : LoadTime.js
# Preferences : -
# Version : 0.5 2008-10-05
# --------------------------------------------------------------------------------

LoadTime_RunCode {
setpref( BOOL, "k-meleon_loadtime_active", TRUE);
$_LoadTime_URL = $URLBAR ;
#$_LoadTime_URL = prompt("Enter or paste here the URL for the page to test.","Website URL");
### ORG exec("wscript.exe \"".$_LoadTime_Path."\" 1");
exec("wscript.exe \"".$_LoadTime_Path."\" 1 ".$URLBAR);
open( $_LoadTime_URL );
}

_LoadTime_Loaded_Run {
getpref( BOOL, "k-meleon_loadtime_active" )== true ? &_LoadTime_Loaded_RunTrue : 0;
}

_LoadTime_Loaded_RunTrue {
exec("wscript.exe \"".$_LoadTime_Path."\" 2 ".$URLBAR) ;
togglepref( BOOL, "k-meleon_loadtime_active") ;
statusbar("TEST IS COMPLETE");
}
_LoadTime_BuildMenu{
setmenu("&Tools",macro,"LoadTime Test",LoadTime_RunCode,Misc);
}

_LoadTime_GetPath{
### LoadTime.js is supposed to be located in the same folder as LoadTime.kmm (this file).
### If it isn't, Windows Script Host will notify the user.
### Double underscore makes variable local only.
$__Data=readfile(getfolder(MacroFolder)."\\LoadTime.kmm");

$_LoadTime_Path=getfolder($__Data==""?UserMacroFolder:MacroFolder)."\\LoadTime.js";
##$_LoadTime_Path=getfolder($__Data==""?UserMacroFolder:MacroFolder)."\\test.js";
}

$OnLoad=$OnLoad."_LoadTime_Loaded_Run;";
$OnStartup=$OnStartup."_LoadTime_GetPath;";
$OnInit=$OnInit."_LoadTime_BuildMenu;";
$macroModules=$macroModules."LoadTime;";


THE JSCRIPT::: LoadTime.js

var d, t, pl;

//~ if(WScript.Arguments.count()==1)
if(WScript.Arguments.count()==2)
{
//~ WScript.Echo (WScript.Arguments(1));
var WshShell = new ActiveXObject("WScript.Shell")
if (WScript.Arguments(0) == 1)
{
d = new Date();
t = d.getTime();
WshShell.RegWrite ("HKCU\\Software\\KMELEON\\LoadTime\\start", t, "REG_SZ");
}
if (WScript.Arguments(0) == 2)
{
d = new Date();
t = d.getTime();
WshShell.RegWrite ("HKCU\\Software\\KMELEON\\LoadTime\\finish", t, "REG_SZ");
var $_LoadTime_Start = WshShell.RegRead ("HKCU\\Software\\KMELEON\\LoadTime\\start") ;
var $_LoadTime_Finish = WshShell.RegRead ("HKCU\\Software\\KMELEON\\LoadTime\\finish") ;
var $_LoadTime_Elapsed = $_LoadTime_Finish - $_LoadTime_Start ;
var $_LoadTime_Seconds = $_LoadTime_Elapsed / 1000 ;
//~ WScript.Echo( $_LoadTime_Seconds + " Seconds elapsed time") ;

var fso = new ActiveXObject("Scripting.FileSystemObject");
var a = fso.OpenTextFile("c:\\KM_SpeedTest.txt",8);
a.WriteLine( (WScript.Arguments(1)) + " --> " + $_LoadTime_Seconds + " Seconds elapsed time");
a.Close();

WshShell.RegDelete ("HKCU\\Software\\KMELEON\\LoadTime\\start") ;
WshShell.RegDelete ("HKCU\\Software\\KMELEON\\LoadTime\\finish") ;
}
}
else
{

objArgs = WScript.Arguments ;
WScript.Echo(WScript.Arguments.Count());
for (i=0; i<objArgs.length; i++)
{
WScript.Echo(objArgs(i))
}
}

MrWoo.

Oops. The only thing I forgot to do was verify the existence of the log file, since I open the fso in append mode. For now just make c:\KM_SpeedTest.txt. Or you can change the path in the script as well.

MrWoo.

Edit: Oh yeah, lol, it took me awhile to figure out that in a .kmm file, the period (.) is the concat operator. Duh.



Edited 1 time(s). Last edit at 10/08/2008 07:50AM by MrWoo.

Quote
MrWoo
Next I think I will figure out how to either open the logfile so you can create a divider and show your tweaks, or maybe some kind of prompt.

Some of code that I had planned for the prefs stuff will work for this. Please post a list of the most important prefs that you are changing for tests. It does not have to be complete. I just need an idea of what I need to work with. I plan to put them in a plain text file. That way you can add or substract prefs from the report as you desire. I don't need the value of the pref. I just need the name and the type. Like this:

INT, content.notify.backoffcount
INT, content.notify.interval
BOOL, content.notify.ontimer
INT, content.notify.threshold
STRING, image.animation_mode

I think I can have the macro/js provide to the log file the prefs with their value at time of the test.

I will try to upgrade my comments also.

I will do that tonight. I have a couple questions.

First, when you exec() say wscript, does the macro wait for a return value at all? Or does it just proceed?

Second, if you had a text file with URL's you can read it and I presume do enough string manip to have example 5 different URL's. Then programatically run the macro on each.

Or, if you had say no cache at all, and had 5 tabs open, would it be possible for the macro to send first parm to jscipt, write the reg val, then reload each, and on final tab loading then send second parm to jscript for reg compare and logging?

MrWoo

The exec statement is not modal. It runs whatever is put in there. It could be WScript.exe or Notepad.exe. The next statement following the exec statement is run imediately. Statements like alert, confirm, and prompt are modal. They wait for the user's input. Following statements do not run until the input is received. The only way the exec statement appears as modal is when it is the last statement in a module.

EXAMPLE ( note that this is finished and works)
## Run notepad.exe on the pref_watch file
_LoadTime_EditPrefs{
exec("notepad.exe \"".$_LoadTime_Prefs_Path."\"") ;
}

I am planning to have a file where you can place several URLs. That file will be read and each URL will be tested. Likely will have another place where the number of tests to run on every URL will be stored. That is small so it may be a user_pref.

If I ever get finished the menu will have these items, I think.

Run the test
Setup the prefs list ( not edit, but prepare for transfer from js to kmm)
Edit the prefs list
Edit the URLs list
Edit the number of tests

Goodness, this thing just keeps growing. I cannot give you a timeline yet. Parts are finished, but they in separate pieces and have to be put together. URLs list and number of test are not even started yet.

Actually these settings are a completely waste of time.

guenter Wrote:
-------------------------------------------------------
> All Gecko's use more or less the same speed
> twweaks. Here is an old list.
> AFAIK It is also in Forum history.
>
> Excerpts from Firefox tweak guide created by
> Achilles
>
> Most of us have heard of these tweaks, and they
> can be used with any Mozilla.
> The settings are inserted into user.js, which
> feeds them into prefs.js on start.
> Achilles based his summary at Tweakfactor on a
> thread at Mozilla Forum.
>
> K-Meleon and some other Gecko browsers that use
> their native OS widgets will not use tweaks for
> XUL items.
> - So not all tweaks are applicable for k-m, but
> even those that are not usefull do no harm; k-m
> just overlooks them.
> Since K-Meleon goes with low resources - Slow
> Computer means slower than Pentium 500 for us ;-)
> .
> Quick and Dirty Settings
>
> user_pref("network.http.pipelining", true);
> user_pref("network.http.pipelining.firstrequest",
> true);
> user_pref("network.http.pipelining.maxrequests",
> 8);
> user_pref("nglayout.initialpaint.delay", 0);
>
> Pipelining does multiple data requests at once and
> should speed things up. Initial Paint Delay
> actually slows down the rendering of the ENTIRE
> page but I tend to start reading before the entire
> page is rendered ; - ) this gives me the
> impression that the page loads faster.
> Common to all configurations
>
> These are the settings that seem to be common to
> all configuration files regardless of connection
> speed or computer speed with a couple of additions
> - plugin paths can be found with about:plugins and
> the bookmark menu delay is turned off.
>
> user_pref("network.http.pipelining", true);
> user_pref("network.http.proxy.pipelining", true);
> user_pref("network.http.pipelining.maxrequests",
> 8);
> user_pref("content.notify.backoffcount", 5);
> user_pref("plugin.expose_full_path", true);
> user_pref("ui.submenuDelay", 0);
>
> Fast Computer Fast Connection
>
> user_pref("content.interrupt.parsing", true);
> user_pref("content.max.tokenizing.time",
> 2250000);
> user_pref("content.notify.interval", 750000);
> user_pref("content.notify.ontimer", true);
> user_pref("content.switch.threshold", 750000);
> user_pref("nglayout.initialpaint.delay", 0);
> user_pref("network.http.max-connections", 48);
> user_pref("network.http.max-connections-per-server
> ", 16);
> user_pref("network.http.max-persistent-connections
> -per-proxy", 16);
> user_pref("network.http.max-persistent-connections
> -per-server", 8);
> user_pref("browser.cache.memory.capacity",
> 65536);
>
> A couple settings of note - Firefox is allocated
> 4096 KB of memory by default and in this
> configuration we give it roughly 65MB as denoted
> by the last line. This can be changed according to
> what is used.
>
> Fast Computer, Slower Connection
>
> This configuration is more suited to people
> without ultra fast connections. We are not talking
> about dial up connections but slower DSL / Cable
> connections.
>
> user_pref("content.max.tokenizing.time",
> 2250000);
> user_pref("content.notify.interval", 750000);
> user_pref("content.notify.ontimer", true);
> user_pref("content.switch.threshold", 750000);
> user_pref("network.http.max-connections", 48);
> user_pref("network.http.max-connections-per-server
> ", 16);
> user_pref("network.http.max-persistent-connections
> -per-proxy", 16);
> user_pref("network.http.max-persistent-connections
> -per-server", 8);
> user_pref("nglayout.initialpaint.delay", 0);
> user_pref("browser.cache.memory.capacity",
> 65536);
>
> Fast Computer, Slow Connection(dial-up)
> user_pref("browser.xul.error_pages.enabled",
> true);
> user_pref("content.interrupt.parsing", true);
> user_pref("content.max.tokenizing.time",
> 3000000);
> user_pref("content.maxtextrun", 8191);
> user_pref("content.notify.interval", 750000);
> user_pref("content.notify.ontimer", true);
> user_pref("content.switch.threshold", 750000);
> user_pref("network.http.max-connections", 32);
> user_pref("network.http.max-connections-per-server
> ", 8);
> user_pref("network.http.max-persistent-connections
> -per-proxy", 8);
> user_pref("network.http.max-persistent-connections
> -per-server", 4);
> user_pref("nglayout.initialpaint.delay", 0);
> user_pref("browser.cache.memory.capacity",
> 65536);
>
> Slow Computer, Fast Connection
> user_pref("content.max.tokenizing.time",
> 3000000);
> user_pref("content.notify.backoffcount", 5);
> user_pref("content.notify.interval", 1000000);
> user_pref("content.notify.ontimer", true);
> user_pref("content.switch.threshold", 1000000);
> user_pref("content.maxtextrun", 4095);
> user_pref("nglayout.initialpaint.delay", 1000);
> user_pref("network.http.max-connections", 48);
> user_pref("network.http.max-connections-per-server
> ", 16);
> user_pref("network.http.max-persistent-connections
> -per-proxy", 16);
> user_pref("network.http.max-persistent-connections
> -per-server", 8);
> user_pref("dom.disable_window_status_change",
> true);
>
> One of the changes made for this particular
> configuration is the final line where the status
> bar is disabled for changing web pages to save
> processor time.
>
> Slow Computer, Slow Connection(dial-up)
>
> user_pref("content.max.tokenizing.time",
> 2250000);
> user_pref("content.notify.interval", 750000);
> user_pref("content.notify.ontimer", true);
> user_pref("content.switch.threshold", 750000);
> user_pref("nglayout.initialpaint.delay", 750);
> user_pref("network.http.max-connections", 32);
> user_pref("network.http.max-connections-per-server
> ", 8);
> user_pref("network.http.max-persistent-connections
> -per-proxy", 8);
> user_pref("network.http.max-persistent-connections
> -per-server", 4);
> user_pref("dom.disable_window_status_change",
> true);

IMHO these settings are a urban legend. In fact, Gecko engine is getting worse and worse and IMHO Gecko is slower. No tricks or tweaks can give the real speed that others engine have since ever. Event IE trident browser or KHTML, or Krestel in Opera. Why don't you think that if tweaking K-Meleon, Flock or Firefox will have more speed itself? Why tweaking? Tweaking Gecko is a shame. Read this:

http://mso-chronicles.blogspot.com/2008/07/gecko-is-getting-worse.html

Well, I have been using Opera for a very long time now. It is still fast, but not any faster to the eye than KM IMO. This is the reason for wanting to speed it up. Already the two are so close that I cannot visually tell the difference. IMO Opera is becoming mainstream bloatware. It becomes everything to everyone. For myself, I always enjoyed the obscurity of a very fast browser that not many people used. Good for them that they have found success, and hopefully they make some cash as well. But I like things less the extra stuff that I never use. KM provided that for a long time, and now I have found the last couple versions to be feature rich enough at default to draw my attention from Opera.

I think FF is junk. I cannot stand it, even to the point of using IE. I don't know why KM is so much more responsive and I don't really care. I am not into following what is 'the best' according to the masses. I am into the app that affords me either ridiculous speed and low footprints or something that is very basic but that I can customize.

Not trying to have any attitude here, please don't take it that way at all. I just happen to have a twisted view on what I want in software is all.

And I look at it like this, even if after using these tweaks, I find only a 1 or 2 second difference, either way, it is not wasted time. True I may not be able to tell the difference, but I sure will be able to get a better understanding of something that I did not know before. That alone is worht it to me.

I wonder if peoples opinions that the gecko engine is getting slow is related to machine specific settings. Or network providers, or even just what they are looking at online.

One more note, even though I still greatly like Opera, but don't use it, there was one issue that always bugged me. I use Outpost firewall (older version only), and Opera or FF would periodically give me a blue screen if that firewall was running. It always happened because of the cache. Too many tabs open for too long a time, or too many bits downloaded would do it. It was reproducable on many machines with different versions of the browsers and firewall. KM never did that. Perhaps due to the layers rather than the tabs, I don't know. But I can say that with IE/Opera/FF I have had bsod of some kind at some point. KM so far as my time ever using it is concerned has never bsod on me.

Thanks for the information. One should at least look at the other side of the coin to develop a better view.

MrWoo.

MrWoo Wrote:
-------------------------------------------------------
> Well, I have been using Opera for a very long time
> now. It is still fast, but not any faster to the
> eye than KM IMO. This is the reason for wanting to
> speed it up. Already the two are so close that I
> cannot visually tell the difference. IMO Opera is
> becoming mainstream bloatware. It becomes
> everything to everyone. For myself, I always
> enjoyed the obscurity of a very fast browser that
> not many people used. Good for them that they have
> found success, and hopefully they make some cash
> as well. But I like things less the extra stuff
> that I never use. KM provided that for a long
> time, and now I have found the last couple
> versions to be feature rich enough at default to
> draw my attention from Opera.
>
> I think FF is junk. I cannot stand it, even to the
> point of using IE. I don't know why KM is so much
> more responsive and I don't really care. I am not
> into following what is 'the best' according to the
> masses. I am into the app that affords me either
> ridiculous speed and low footprints or something
> that is very basic but that I can customize.
>
> Not trying to have any attitude here, please don't
> take it that way at all. I just happen to have a
> twisted view on what I want in software is all.
>
> And I look at it like this, even if after using
> these tweaks, I find only a 1 or 2 second
> difference, either way, it is not wasted time.
> True I may not be able to tell the difference, but
> I sure will be able to get a better understanding
> of something that I did not know before. That
> alone is worht it to me.
>
> I wonder if peoples opinions that the gecko engine
> is getting slow is related to machine specific
> settings. Or network providers, or even just what
> they are looking at online.
>
> One more note, even though I still greatly like
> Opera, but don't use it, there was one issue that
> always bugged me. I use Outpost firewall (older
> version only), and Opera or FF would periodically
> give me a blue screen if that firewall was
> running. It always happened because of the cache.
> Too many tabs open for too long a time, or too
> many bits downloaded would do it. It was
> reproducable on many machines with different
> versions of the browsers and firewall. KM never
> did that. Perhaps due to the layers rather than
> the tabs, I don't know. But I can say that with
> IE/Opera/FF I have had bsod of some kind at some
> point. KM so far as my time ever using it is
> concerned has never bsod on me.
>
> Thanks for the information. One should at least
> look at the other side of the coin to develop a
> better view.
>
> MrWoo.


Sorry but I think you didn't unsderstand me.
I only want to ask yourself why do you have to make the speed tweaking experiments? Wich is the reason? What's the meeting point of the conclusions?
Isn't there a development team to probe all this facts?
The final user ONLY be able to use the final product, not tweaking.
If the development team knows how to increase speed, that's not your problem.
IMHO K-Meleon is a good product, but HAO's K-Meleon has reached more speed.
Why these differences between the official version and unofficial versions.
That's I wanted to say. Try to make these experimentes with HAO's K-Meleon.
But please, one or two seconds of speed difference don't make the choice.
Mr. Wilson.

c. wilson schrieb:
-------------------------------------------------------
> Which is the reason? What's the meeting point of the conclusions?

Easy - we do it to get more for our money while we are in the net with this browser. The meeting point is that PPL tested - I myself remember begining for 0.8.2 versions compiled by Monkey-Sage, Ulf and later the fastest private ones by Dorian.

This tweaking is especially interesting when You do not come frome the Urban centers of Spain or Germany but from rural areas of huge states e.g. from Mantana in the Rockies. When all is smooth and Your DSL at normal speed we will never care to find out faster solutions. But when my DSL is down to modem speed (right in Hannover near the center of Europe!) = I start to search solutions for just a little more speed.


c. wilson schrieb:
-------------------------------------------------------

>
> IMHO these settings are a urban legend.

IMHO "legendary" is the word

The tweaks give geckos extra edges (in role plays You must pay for each extra edge) because modern Servers and OS support the tweaks while most browsers do not know them yet. Even a slow Gecko that only competes with itself and has no merrits except a slow start and a superb extension system benefits from such tweaks
The tweaaks give You an edge again untweaked Geckos and all other browsers. Hm let me think - it is like overtaking the neighbor that has exactly the same car. It is not essential but sure nice

I have to repeat some info from this Forum from really old Posts.
1.) Humans need 25% difference to notice different speeds.
2.) 10% faster speed is unnoticable but gives You an extra hour surfing every 10 hours

Anything below 1.) must be somehow messured.
And 2.) happens whether You notice and messure it or not. It gives us more content on a flatrate or more for our Euro/$ if we pay per minute.


>In fact,
> Gecko engine is getting worse and worse and IMHO
> Gecko is slower.

No. I keep a close watch since K-Meleon 0.8. That is several Geckos ago. ( private and official 0.8.x existed with GRE 1.5 - 1.8 & I have seen 0.9.x & 1.x with 3 GRE each). Each Gecko since 1.5 was a little faster than the previous, some where much faster. Yet sometimes other browsers managed to catch up or were even faster for some aspects.



> No tricks or tweaks can give the
> real speed that others engine have since ever.

I do not understand this.
You address K-Meleon Forum.
K-Meleon is currently the fastest Windows browser. And it has been IMHO for 0.6-0.8 & since 1.x when Dorian took it, till now. Including unofficial versions gives us extra. We still tweak K-Meleon since we get even more then!

> Event IE trident browser or KHTML, or Krestel in
> Opera.

? compare List of Engines. When was Kestrel a HTML engine? Without being sure: Kestrel is a JS engine perhaps?

Trident was fast until IE 5.01 and that is long ago.

Recent Operas' oustanding feature was linearB (= the JavaEngine & currently the Opera JS engine is named after the Norsk rune Alphabet ). The capabilies of its html engine were never outstanding (not after Gecko learned fast backward anyway - before that the later feature was more than compensated by the tweaks listed above,and they were adopted by K-Meleon community just when Opera had that fast back and the feature was only experimental for GRE 1.8 ).

> Why don't you think that if tweaking
> K-Meleon, Flock or Firefox will have more speed
> itself? Why tweaking? Tweaking Gecko is a shame.
> Read this:
>
> http://mso-chronicles.blogspot.com/2008/07/gecko-i
> s-getting-worse.html


This blogspot rant in and for Midori a GTK intergrated web-browser is pathetic.
My experience since K-Meleon 0.6 ( that is before Firefox, Flock ... Midori projects even started ) tells me that early versions with few features often have better speed than the final version, which get even slower when all it bugs get fixed (seen the later several times for Opera).

BTW Galeon/Epiphany which are also a GTK intergrated browsers are also faster than e.g. Flock and Firefox under the respective Nix. And K-Meleon is just as integrated into Windows native widgets as they are into GTK.



Edited 2 time(s). Last edit at 10/10/2008 09:20PM by guenter.

-----------------------
Sorry but I think you didn't unsderstand me.
I only want to ask yourself why do you have to make the speed tweaking experiments? Wich is the reason? What's the meeting point of the conclusions?
Isn't there a development team to probe all this facts?
The final user ONLY be able to use the final product, not tweaking.
If the development team knows how to increase speed, that's not your problem.
IMHO K-Meleon is a good product, but HAO's K-Meleon has reached more speed.
Why these differences between the official version and unofficial versions.
That's I wanted to say. Try to make these experimentes with HAO's K-Meleon.
But please, one or two seconds of speed difference don't make the choice.
Mr. Wilson.
-----------------------
I see now. I agree, one or two seconds is not what I am after. In fact, I am not even sad to see no differences, only to try. If some time spent learning something new can for me, increase page loads etc, then I am double lucky. Because I learned and also recieved better performance.

And I would agree that the development team probably has tweaked it. But as is always the case, it is tweaked for an average system. Each system being different, there is always possibles that more can be achieved on my system.

Thank you for clarifications.

MrWoo.

Guenter, how do you know? I in fact do live in the mountains in NW Montana. That is a most strange example considering.

I would say TRUE. That early versions with less features are very much IMO better/faster. In many cases not just browsers.

TRUE as well what you say about Opera, the very reason that I am not as impressed as long ago.

TRUE that tweaking may need to achieve higher % to be seen, but just to know sometimes that 3% faster is still 3% faster is enough when you only want to see what happens.

Good thoughts Guenter.

MrWoo.

No coincidance - Montana is an example I knew from tales from Zeronero from this forum. He told me that service is handicaped by the long distances in Montana.

c. wilson & Spain is also explained. Spain has often good DSL conections.
& I am moderator & can (sometimes) see the cable providers of posters.

p.s. I also used Opera long ago. But I liked k-meleon's speed and efficient use of resources. If K-Meleon did not exist I would probably use Opera (but without Mail and other clients - which IMHO only help to slow it). AFAIK out of box Opera makes efficient use of modem conections - so for some cconections You tweak the Gecko or use other browsers.



Edited 3 time(s). Last edit at 10/11/2008 01:09PM by guenter.