Low memory full feature browsers [Solved]

[Wishes]

Thanks for your work Durhammer (did not read the thread yet). I also think browsers need aaalottttt of mem ...bloat

[lars_the_bear]

EDIT: Okay, it's more than a little clunky. Is JavaScript support HTML5? Whatever it is, Links doesn't seem to support it. THIS site requires it. Probably so does say Google News. Not so much utility as futility. I'm sure it's enjoyed for specific uses, but it doesn't fall under "full featured." Thanks for the suggestion, though!

Good grief: I wasn't actually suggesting that anybody should use links :) My point was only that it's possible to implement a web browser in low memory, if you're willing to sacrifice enough common functionality. And use a very old-fashioned user interface. The more contemporary stuff you include -- JavaScript, HTML5, media support, X, Wayland -- the more bloated the browser will be.

I don't think modern browsers are designed to be efficient in their use of resources but, when you consider what they are called on to do, ~200Mb of RAM doesn't seem all that unreasonable to me. There's a kind-of spectrum of bloat, from full-fat Chromium at one end, to 'links' at the other. All we can do is pick the best compromise we can, between features and bloat, from the small number of browsers that are available.

The real culprit here are the website designers -- I would un-invent JavaScript if I could, but that ship has already sailed. Still, designers -- and the tools they use -- care little for the users of resource-constrained computers.

BR, Lars.

[Durhammer]

No worries, @lars_the_bear !! I just figured, what the hey, let's see just what it can do. Totally agree with you on the business of web page design -- much of it now dedicated to monetizing the web in one way or another (ads, paywalls, what have you). Maybe if someone designed a search engine or news feed/reader that only delivered HTML4 sites, minimalist browsers would be great, but there probably wouldn't be much to see. Sigh.... Anyway, I actually did enjoy trying out these various browsers in my quest. Kinda eye-opening.

[lars_the_bear]

Maybe if someone designed a search engine or news feed/reader that only delivered HTML4 sites, minimalist browsers would be great, but there probably wouldn't be much to see. Sigh....

That's the problem with initiatives like the Gemini protocol. Yes, there's a 'small web' movement, and we can have a simple protocol and lightweight sites, but there's hardly any content. It isn't just because mainstream websites are full of advertising cruft, although many are. It's not cost-effective any more to crank out a large site by hacking on HTML and what-not; site designers and content management systems mostly generate bloated, tech-heavy sites, because it's easier for them to do this than to do something lighter.

As somebody who remembers the web before it became an advertising platform, I have to say that I find this disappointing. But that's progress, I guess :/

BR, Lars.

[Durhammer]

Yeah, me too, Lars. For the record, my foray into computing started ~50 years ago at an outfit in the Research Triangle Park in NC, where we hosted the "Triangle Universities' Computation Center" (TUCC), which had a huge IBM 360 mainframe with 200K magnetic cores memory and row after row of IBM tape drives. Woo! Your average low-end cell phone has more computing power than that. Those were the days! No email, before usenet even, and certainly no WWW (and no Microsoft, no Windows, no viruses, and no Linux!). We were an "information retrieval" service for research; people either would call in or snail-mail requests, engineers would run queries against various specialized databases, and would snail-mail back reports to the requestors. As someone now famous says, "we're not going back"! :-)

[rokytnji.1]

Finally got on here on 32 bit Midori. You are welcome to test out 64 bit I guess.

https://astian.org/midori-browser/download/linux/

[Durhammer]

Thanks, @rokytnji.1 -- I had tried Midori on my main Laptop (MX-21, Xfce) and did not have a convincing experience with it -- it might have been from leftover settings from a previous install, I dunno, but it just did weird things there. HERE, it behaves much better (I'm writing this in Midori). Still, I decided to do a more or less head-to-head comparison with Brave, which I continually am surprised by. After going through the Midori setup, and adding the same two extensions I use on Brave -- Bitwarden and UBO -- I copied/pasted URLs from Brave into Midori, and refreshed each tab on each browser, then started my ps_mem.py tests. I just re-ran the command somewhat randomly over a period of 30-40 minutes. I did NOT touch either browser in the meantime, letting each just do whatever they do in the background. It's VERY interesting to see how the memory use goes up, goes down, for each, but in the end, it appears that Brave just does a bit more to "get out of the way" than does Midori. I find it both weird and compelling. Initially, Brave took more RAM than Midori. Then, over time, doing nothing with either, in general Brave shrank and Midori grew. WEIRD!

Code: Select all

322.5 MiB +  41.7 MiB = 364.2 MiB	midori (12)
414.9 MiB +  79.5 MiB = 494.4 MiB	brave (17)
---------------------------------
329.7 MiB +  42.2 MiB = 371.8 MiB	midori (12)
403.6 MiB +  81.0 MiB = 484.5 MiB	brave (17)
---------------------------------
330.7 MiB +  42.2 MiB = 372.8 MiB	midori (12)
404.4 MiB +  81.2 MiB = 485.6 MiB	brave (17)
---------------------------------
331.6 MiB +  43.5 MiB = 375.1 MiB	midori (12)
402.8 MiB +  80.2 MiB = 482.9 MiB	brave (17)
---------------------------------
326.2 MiB +  53.9 MiB = 380.0 MiB	brave (17)
419.2 MiB +  87.4 MiB = 506.6 MiB	midori (15)
---------------------------------
349.0 MiB +  86.2 MiB = 435.2 MiB	brave (17)
430.5 MiB +  69.6 MiB = 500.1 MiB	midori (15)
---------------------------------
341.9 MiB +  85.8 MiB = 427.6 MiB	brave (17)
429.6 MiB +  84.2 MiB = 513.8 MiB	midori (15)
---------------------------------
346.5 MiB +  85.9 MiB = 432.4 MiB	brave (17)
426.7 MiB +  83.9 MiB = 510.6 MiB	midori (15)
---------------------------------
345.6 MiB +  86.8 MiB = 432.4 MiB	brave (17)
432.9 MiB +  84.0 MiB = 516.9 MiB	midori (15)
---------------------------------
341.6 MiB +  89.4 MiB = 431.0 MiB	brave (17)
433.2 MiB +  84.2 MiB = 517.4 MiB	midori (15)
---------------------------------
345.9 MiB +  96.5 MiB = 442.5 MiB	brave (18)
449.9 MiB +  85.5 MiB = 535.4 MiB	midori (15)
---------------------------------
332.9 MiB +  87.0 MiB = 419.9 MiB	brave (17)
450.6 MiB +  85.8 MiB = 536.4 MiB	midori (15)
---------------------------------
271.2 MiB +  62.8 MiB = 334.0 MiB	brave (17)
528.8 MiB +  53.9 MiB = 582.7 MiB	midori (15)
---------------------------------
285.9 MiB +  71.5 MiB = 357.4 MiB	brave (17)
435.4 MiB +  71.9 MiB = 507.3 MiB	midori (15)
---------------------------------
287.7 MiB +  71.7 MiB = 359.4 MiB	brave (17)
462.3 MiB +  87.1 MiB = 549.3 MiB	midori (15)
---------------------------------
291.4 MiB +  72.4 MiB = 363.8 MiB	brave (17)
433.8 MiB +  87.1 MiB = 520.9 MiB	midori (15)
---------------------------------
289.2 MiB +  72.7 MiB = 361.9 MiB	brave (17)
432.3 MiB +  87.0 MiB = 519.3 MiB	midori (15)
---------------------------------
291.9 MiB +  72.9 MiB = 364.7 MiB	brave (17)
431.2 MiB +  87.2 MiB = 518.4 MiB	midori (15)

[/df]

One word: NoScript

With the appropriate latest versions of the add-on installed ,,,

Firefox 128 ESR 6 tabs:

Code: Select all

287.9 MiB +  28.4 MiB = 316.4 MiB	firefox-esr (11)

SeaMonkey 2.53.19 48 tabs (yes, I know), 1 Mail window, 1 Mail message window

Code: Select all

421.8 MiB + -13374.5 KiB = 408.8 MiB	seamonkey

The reason for using SM is not just that it's generally a low memory browser but because it's the only platform for classic Mozilla add-ons that make browsing and email bearable. The reason for using FF is that it supports NoScript, and is not Chrome.

FWIW ps_mem.py is included in the MX-Goodies package.

[Durhammer]

Thanks for the suggestions, @/df ! I'll have to find and try out FF ESR (I think it was default in some distro I tested under VBox). Earlier in this post, I had already dismissed Seamonkey. Sorry!

Meanwhile, I installed NoScript on both Midori and Brave. I do like the fact that NoScript's initial installation (saying that because I think all of them have to store somewhere what sites are accepted and what aren't, right?) is a mere 943 KB, whereas UBO is 4 MB *initially* in Midori, and whereas in Brave, UBO is not offerred, but UBO Lite is, and it is taking up 35.1 MB right now, and NoScript is taking up 2.3 MB. Quite a considerable savings or memory right there! I don't know how to tell how much the extensions are taking up currently in Midori -- all I get is the initial extension size.

And yet -- and yet -- I figured this would be a great chance to test these as ad blockers, 'cause otherwise, why have them at all?

So with the current two browsers standing (for now!) -- Brave and Midori -- I think Brave is Chromium based and Midori Firefox, I turned off any ad-blocking or script-blocking and ran a baseline test on "adblock-tester.com" . Midori has UBlockOrigin (original) and NoScript installed (ignoring Bitwarden installed everywhere). Brave has NoScript and UBlockOrigin Lite installed.

Baseline test results (test is a score of x out of 100 for adverts and such blocked by the ad blocker), NO ad-blocking at all -- oddly enough-- 39/100.

In *either* Midori or Brave, without any other ad-blocking, NoScript earned a score of 65/100.

Strangely, in Brave, adding Brave's "Shields Up" seems to have been stifled by NoScript being active -- no better score (still 65/100). Strange.

UBlockOrigin (the full monty, only on Midori), earned a 100/100 score. Rah!

On Brave, other than the NoScript results as above, UBlockOrigin "Lite" all by itself scored 91/100.

With both NoScript and UblockOrigin Lite *disabled*, and Brave's "Shields Up", it got a score of 96/100. Nice!

Both UBO and UBOL, and Brave all have script blocking, so I don't thing NoScript adds much -- especially given the fact that by itself, Brave blocks more than when in combination with NoScript or UBOL.

So what does all this mean? Well, for ME, it would seem that all I need is Brave. No NoScript, no UBOL. It means I can save a little more memory that is just taken up by the extension itself. I'll be giving it a try.

A final note: I don't know why, but Midori is slow as molasses. It's on the way out, ready for a trial of FF ESR. More later!

[Durhammer]

A side note about Ad-blockers -- for one thing, the handwriting is on the wall for UBO's demise. Raymond Hill probably only will keep UBlockOrigin Lite going. My favorite Ad-blocker is Adguard, with its near-perfect ad blocking and its cosmetic massaging, but it's way too big for this 2 GB RAM machine.