AquaWeb micro - A Stupidly Fast Minibrowser for PPC OS X

[Dronecatcher]

Hello everybody, I am Looking4awayout of L4Soft, the creator of the UOC Patch.

Thanks for highlighting some of the problems here - to avoid any confusion could you provide a link to the correct versions?
I tried this out when it was first referenced in the forum and the prefs file resulted is some bizarre behaviour for TenFourFox - I had to reinstall it to get things back to normal.

 

[looking4awayout]

https://msfn.org/board/topic/178306...wsers-for-old-machines-now-for-macintosh-too/ This is where you can get the UOC Patch and the Enforcer.

You can get the Macintosh version here: https://www.mediafire.com/file/wf1qdc98fr6sk43/UOC_Patch_Mac_N2H.zip/file

And its Enforcer here:

UOC_Enforcer_38ESR_N2H

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

Please remember that the Macintosh version has been conceived for Arctic Fox, as it is built on the codebase of Firefox 38 ESR. Only one person besides Sparty411 beta tested it on TenFourFox and on his machine, a Powerbook G4, seemed to perform well enough.

Installing the patch on the Macintosh is as simple as doing it on Windows. You just unzip the patch and put the UOC_Patch_Mac.js file into the following folder:

Applications\[Browser folder]\Contents\Resources\Defaults\Pref

The UOC Enforcer for 38 ESR browsers is the version that must be used with the Macintosh one. You can install the Enforcer on the Macintosh by putting the user.js file in the following directory:

Macintosh HDD\Users\[Username]\Library\Application Support\[Browser Name]\Profiles\[Profile Name]\

Make sure you back up your prefs.js file before installing the UOC Patch and the Enforcer.

I would be glad if someone further optimizes it to run on PowerPC Macintoshes, as the Windows version works well enough to optimize old hardware.

It is always recommended to start with a clean profile to avoid potential issues.

 

[Dronecatcher]

I would be glad if someone further optimizes it to run on PowerPC Macintoshes, as the Windows version works well enough to optimize old hardware.

Thanks. I did some benchmarking yesterday:

Low End Leopard Performance

Can you give examples of how you reach this conclusion - which sites perform better? I've done some benchmarks and the Ultimate prefs are slower than my own that are a modified version of the @eyoungren prefs? On latest TFF Powerbook 12" 1.33 Tiger (Time in seconds to load page - two sets)...

forums.macrumors.com

I'll install these prefs and compare.

 

[looking4awayout]

Wonderful, thanks! I hope the instructions are clear enough.
The UOC Patch can be tweaked to suit your machine's needs. While the original Windows one has been based upon the specs of my own computer, an overclocked Pentium III Tualatin, I have easily ported it to the "Macintel" Mac Mini 1.1 by removing everything related to Direct3D, replacing it with OpenGL. On the "Macintel" and at least one Powerbook G4, it works as it should. But I don't know how it runs on other Power Macintoshes.

Please follow the installation instructions carefully. The UOC Patch and the Enforcer are not prefs.js files and must not be installed like a normal prefs.js. I posted the installation instructions above.

 

[Dronecatcher]

These are the result with the ones I did yesterday repeated:

FPR16 19/10/2019 Powerbook 12" 1.33 Tiger

Standard
Launch to Home (Google) 15.33 12.17
MacRumors PPC 14.02 18.45
Youtube Video 37.99 40.86

Personal
Launch to Home (Google) 11.79 11.86
MacRumors PPC 7.27 6.54
Youtube Video 25.78 28.08

Ultimate
Launch to Home (Google) 19.40 13.88
MacRumors PPC 13.39 13.48
Youtube Video 29.09 28.91

UOC
Launch to Home (Google) 11.93 11.94
MacRumors PPC 15.15 12.84
Youtube Video 30.12 27.55
[automerge]1571583313[/automerge]
Maybe a combo of the UOC prefs and my own might yield better results?

 

[looking4awayout]

You can try.

You might have to tweak the paint delay settings to see if it loads the page faster. The way I set it is to load pages without stressing the CPU too much, because at least on my Tualatin machine, I noticed that the shorter the loading times (decreasing the paint delay, increasing the maximum amount of connections, as well as the maximum pipelined connections), the more it takes a toll on the CPU. So it makes it a little bit faster, but without being too CPU heavy.

Try to edit the maximum amount of connections, reduce the paint delay and increase the amount of web socket and pipelined connections. It might make a difference but your CPU usage might increase when a page loads.

 

[Dronecatcher]

You might have to tweak the paint delay settings to see if it loads the page faster. The way I set it is to load pages without stressing the CPU too much, because at least on my Tualatin machine, I noticed that the shorter the loading times (decreasing the paint delay, increasing the maximum amount of connections, as well as the maximum pipelined connections), takes a toll on the CPU. So it makes it a little bit faster, but without being too CPU heavy.

Try to edit the maximum amount of connections, reduce the paint delay and increase the amount of web socket and pipelined connection. It might make a difference but your CPU usage might increase when a page loads.

Thanks for that insight. Yes, on G4 portables less CPU for longer load is a good tradeoff - particularly on my 12" Powerbook, where the fan takes off regularly with TenFourFox.

 

[looking4awayout]

Speaking of which, how does your Power Macintosh perform with the UOC Patch and the Enforcer? How's scrolling? Is it faster than stock TFF? Please let me know! I long to know how it runs on a real Power Macintosh.

Unfortunately TenFourFox doesn't want to work on my Mac Mini 1.1 even with Rosetta installed so a feedback from a real PowerPC system would be nice to hear. Besides loading times, of course.

 

[Dronecatcher]

Speaking of which, how does your Power Macintosh perform with the UOC Patch and the Enforcer? How's scrolling? Is it faster than stock TFF? Please let me know! I long to know how it runs on a real Power Macintosh.

Unfortunately TenFourFox doesn't want to work on my Mac Mini 1.1 even with Rosetta installed so a feedback from a real PowerPC system would be nice to hear. Besides loading times, of course.

Yes, I did loading times because that was directly measureable, scrolling and clicking through is more subjective.
I'll do some testing later and report back. @z970mp should be able to provide more feedback on actual browsing.

 

[looking4awayout]

Maybe @z970mp can test the UOC Patch for Macintosh and the Enforcer I provided, to see how it behaves.

@Dronecatcher DM sent.

 

[Dronecatcher]

Maybe @z970mp can test the UOC Patch for Macintosh and the Enforcer I provided, to see how it behaves.

@Dronecatcher DM sent.

Check out @eyoungren 's thread, that's the daddy of prefs tweaking - mine's different with a few tweaks borrowed from his.

My TenFourFox About:Config tweaks - and my addons

QUICK LINKS PowerUOC Project. Link to alternative TenFourFox tweak project. Legacy Addon Site

forums.macrumors.com

 

[z970]

These are the result with the ones I did yesterday repeated:

FPR16 19/10/2019 Powerbook 12" 1.33 Tiger

Standard
Launch to Home (Google) 15.33 12.17
MacRumors PPC 14.02 18.45
Youtube Video 37.99 40.86

Personal
Launch to Home (Google) 11.79 11.86
MacRumors PPC 7.27 6.54
Youtube Video 25.78 28.08

Ultimate
Launch to Home (Google) 19.40 13.88
MacRumors PPC 13.39 13.48
Youtube Video 29.09 28.91

UOC
Launch to Home (Google) 11.93 11.94
MacRumors PPC 15.15 12.84
Youtube Video 30.12 27.55
[automerge]1571583313[/automerge]
Maybe a combo of the UOC prefs and my own might yield better results?

Something that occurred to me while scanning your prefs; there's nothing notable I could find that the Ultimate prefs didn't already have (aside from pipelining and max-connections being cut in half, which might warrant investigation), so it's hard to pinpoint why it would be so much faster in MacRumors (I don't use Google; I haven't tested). - Just to be sure, were all addons turned off while conducting these tests? Another thing I had in mind during development was mass image rendering speed. How do each of them compare in rendering a page of images?

Hello everybody, I am Looking4awayout of L4Soft, the creator of the UOC Patch.

Welcome to the forums!

Maybe @z970mp can test the UOC Patch for Macintosh and the Enforcer I provided, to see how it behaves.

Originally, the UOC Patch (for Macintosh and 45 ESR) did not work with TFF, the browser closing itself immediately after opening. So, I looked for the problem preference, removed it, and merged it with our own preferences we've devised over the years, while tweaking a couple things on my own account.

It was also based on the 45 ESR Patch because TenFourFox is FF 45, and I thought it to take better advantage of TFF's newer technologies, as well as being a universally-compatible fork.

Going over these new preferences, it seems that the problem preference is gone, and that the Macintosh version has since changed a couple things as well... Perhaps, a full rewrite is in order...
[automerge]1571594222[/automerge]

 

[looking4awayout]

Yes, I have rewritten the entire Patch and the Enforcer from scratch, as it was beginning to be too much bloated, plus there were some personal preferences buried in the file that were not working as I wanted them to be.

So the new version of the UOC Patch and the Enforcer only contains the essential modifications, slimming it down in size and optimizing the browser more than the older versions did, at least on regular PCs.

On the Macintosh, besides my Mac Mini 1.1, I have no idea.

 

[z970]

Yes, I have rewritten the entire Patch and the Enforcer from scratch, as it was beginning to be too much bloated, plus there were some personal preferences buried in the file that were not working as I wanted them to be.

I agree. I had an inkling that preferences size may contribute to browsing speeds.

At the time of creation of course, your last post was several weeks to a month or so before, so I assumed it was mostly "finished".

On the Macintosh, besides my Mac Mini 1.1, I have no idea.

Now that we're in the same place, we can actually coordinate. Maybe you can take care of the Windows / Linux side, and I'll do the same for the TFF / PPC side.

 

[looking4awayout]

Now that we're in the same place, we can actually coordinate. Maybe you can take care of the Windows / Linux side, and I'll do the same for the TFF / PPC side.

That would be nice. I have developed and tested the patch on the machines I own. But as I don't have a Power Macintosh (and my "Macintel" cannot fit that void anyway!), your help is gold to me.

A PowerUOC Patch would definitely help at making these old machines fare better on the web. The Enforcer should not need any modifications on the other hand, as it only contains the "stubborn" entries that for some reasons, are hardcoded to not be editable with a simple defaults replacement file.

 

[Dronecatcher]

Something that occurred to me while scanning your prefs; there's nothing notable I could find that the Ultimate prefs didn't already have (aside from pipelining and max-connections being cut in half, which might warrant investigation), so it's hard to pinpoint why it would be so much faster in MacRumors (I don't use Google; I haven't tested). - Just to be sure, were all addons turned off while conducting these tests? Another thing I had in mind during development was mass image rendering speed. How do each of them compare in rendering a page of images?

It might not be so much that mine has more but that it doesn't have some of yours that might be detrimental on my Powerbook?
That's right, no add-ons turned on, in fact none installed for this test.
For image rendering, the Youtube page was the only example of that for which yours and my prefs were pretty much the same.

 

[z970]

It might not be so much that mine has more but that it doesn't have some of yours that might be detrimental on my Powerbook?

That's what I might be figuring...

For image rendering, the Youtube page was the only example of that for which yours and my prefs were pretty much the same.

Try searching something, and then click on the search engine Images tab. Tons of images all at once.

 

[looking4awayout]

@Dronecatcher @sparty411 and @z970mp I have done some other updates to the UOC Patch, including the Macintosh version. Please update to the newest version! I've also updated the 38 ESR version of the UOC Enforcer.

I have applied four parameters from Dronecatcher's optimized prefs.js. It seems it might have improved the loading times of a webpage slightly, but I am not noticing any negative impact on performance, so I kept those parameters.

EDIT: I had to revert two of the four parameters, as they have worsened scrolling performance while pictures are loading.

 

[z970]

@Dronecatcher @sparty411 and @z970mp I have done some other updates to the UOC Patch, including the Macintosh version. Please update to the newest version! I've also updated the 38 ESR version of the UOC Enforcer.

I have applied four parameters from Dronecatcher's optimized prefs.js. It seems it might have improved the loading times of a webpage slightly, but I am not noticing any negative impact on performance, so I kept those parameters.

EDIT: I had to revert two of the four parameters, as they have worsened scrolling performance while pictures are loading.

What were the parameters, out of interest?

 

[looking4awayout]

The two offending parameters were the content max tokenizing time and the content notify interval. Using Dronecatcher's ones caused the browser to stutter to the point it would stop scrolling when loading a page with remotely hosted pictures. Improving the scrolling in a picture heavy web page is one of my main objectives.

The two good ones are used for disabling datareporting and another one that enables the cache on external disk. I'll have to check out the 38 ESR version to see if I have implemented the changes correctly, as I was a bit drowsy when I have implemented them in the patch. ?

 

[z970]

I was a bit drowsy when I have implemented them in the patch. ?

Tell me about it...

 

[looking4awayout]

I'll double check the patch to make sure everything's implemented properly. Meanwhile test it, let me know if there is any improvement.

EDIT: Just investigated the patch. Everything has been implemented correctly. I also have added a parameter that disables the RSS feeds icon on the url bar.

 

[z970]

Here's the working version of the PowerUOC patch, optimized for TenFourFox / Arctic Fox.

In my tests, it renders websites faster than UOC N2H (usually by several seconds), with the help of additional / modified parameters.

Have a go.

 

[looking4awayout]

That's nice. I will inspect the patch and see if I can port those parameters to Windows and see how they work. ?

I am thinking to revert the pipelining settings back to the N2F version as it seems, at least on my Tualatin RDD, that the CPU usage is too excessive when a page loads.

How's scrolling behaving on the Power Macintosh?

 

[z970]

How's scrolling behaving on the Power Macintosh?

Slightly choppier than stock N2H, I believe. However, I think the shortened load / render times make up for it.

FYI, I'm testing on a 1.42 G4.