
Ben Markson
MemberContent Type
Profiles
Forums
Events
Everything posted by Ben Markson
-
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Okay, so that's curious, under Serpent, my example only runs if webgl.disable-angle = true otherwise, like you, I get a black screen and a console error: failed to acquire a renderer - check WebGL or WebGPU is supported. I've long decided that WebGL and XP just don't play well together – flash was vastly superior. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Here is a simple game that uses WEbGL. It's a nice example because it's quick to load and the website gives proper diagnostics for any missing browser features. https://play.ninjadoodle.com/clickplaypics/ Now, can someone tell me how to have it run under XP without maxing out the CPU? For me it idles at 75%. I'd always assumed it's down to limitations of XP – with the same browser under W7 the CPU is negligible. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Many posts ago I suggested blocking certain animation effects: https://msfn.org/board/topic/184051-my-browser-builds-part-4/page/131/#comment-1250502 In particular: animation: none !important; breaks the kind of stuff you are talking about. Because of that I then changed my suggestion: https://msfn.org/board/topic/184051-my-browser-builds-part-4/page/131/#comment-1250506 Since then @roytam1 introduced the prefs: layout.css.animation.enabled layout.css.transition.enabled Which supersede my css tweaks. Is it possible that you have an animation: none in force? Ben. -
Here's a blast from the past that I'm still using. ZoneAlarm. But not the later bloatware versions but back when it was lean and mean... v2.6 Its use as a firewall is mostly redundant as nowadays that kind of stuff is handled at the router, but as an egress filter I still find it very useful. I'm still routinely disappointed how the first thing some 'legitimate' software wants to do is to phone home. I also think it's a good protection against malware that rides on the back of another program and whose job it is to open the door to the fetching of the real nasty. ZoneAlarm will warn when a program you weren't expecting tries to connect to the internet. Ben.
-
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
I changed my mind! It breaks too many things (including MSFN) this seems better: * { animation-iteration-count: 0 !important; } ...it gets rid of the 'infinite' setting (I'm going for a global rule, not a per site rule). Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
I think the main culprit is this: https://b.thumbs.redditmedia.com/vVqCiuMuYjtIEG4538PA7zjsBOXQi37pmfABOgXoEQY.css ...with its CSS Animations. This interests me as I really hate webpages that gratuitously consume CPU after they've rendered. Fair enough if its essential to the function of what's being displayed but not just to add some pointless effects. I came across this as a panacea to killing all CSS Animations: * { /*CSS transitions*/ -o-transition-property: none !important; -moz-transition-property: none !important; -ms-transition-property: none !important; -webkit-transition-property: none !important; transition-property: none !important; /*CSS transforms*/ -o-transform: none !important; -moz-transform: none !important; -ms-transform: none !important; -webkit-transform: none !important; transform: none !important; /*CSS animations*/ -webkit-animation: none !important; -moz-animation: none !important; -o-animation: none !important; -ms-animation: none !important; animation: none !important; } Which I reduced to this: * { transition-property: none !important; transform: none !important; animation: none !important; } It certainly improves that Reddit page. Ben. -
I also have a copy of the source code but, like you, couldn't really find anything useful. As stunnel is stupidly light-weight and easy to deploy (no real install needed) I'm hoping it will keep me going for a few years yet. Anyway, thanks for your comprehensive summary which should be useful to anyone else researching this subject. Jose Alf also deserves a little more prominence for his work. Ben.
-
...time passes... A quick postscript. If using v5.70 there's no need to replace tstunnel.exe, it's only stunnel.exe that causes the system tray bug. The bug was introduced with v5.61 so the v5.60 stunnel.exe is the last working version. The v5.70 stunnel.exe does work but you cannot interact with the running instance, this isn't necessarily a big deal as there's not a huge amount you can do. Stunnel runs and it does its thing without needing to interact with it. If you want to download the exe's from github using just about any browser then you can copy the links from https://github.com/josealf/stunnel-win32 (the left hand column) for example: https://github.com/josealf/stunnel-win32/blob/master/stunnel-win32-5.60-openssl-1.1.1k-installer.exe ...change the 'blob' to 'raw': https://github.com/josealf/stunnel-win32/raw/master/stunnel-win32-5.60-openssl-1.1.1k-installer.exe and it will download. I'd like to find a workaround to the system tray bug but otherwise hopefully this will keep my email working in the face of any 'upgrading' my email provider may perform. Ben.
-
Sorry about that, it's my fault for trying to do a running commentary on something I was still working on – it's what can happen when you get sucked into treating a forum like a chatroom. I removed v5.70, it works (tested with Outlook 2002 email client) but has a weird glitch with not displaying sTunnel's tray icon and no matter what I tried I couldn't get it to appear properly. Went back to v5.60 (there's some indication in the change logs that the system tray code changed with v5.61) and the tray icon is back how it should be. Finally, I took the v5.70 install and replaced just the stunnel.exe and tstunnel.exe files with the v5.60 versions. So, it has the later OpenSSL DLLs but is using an earlier version of stunnel. I'm not sure that this is the best idea but... it works (again tested with Outlook 2002). You can even see the weirdness in the above log (the Compiled versus Running dates). Now, having said all that I use stunnel in a pretty basic way (I am by no means an 'expert' stunnel user) so it is possible that if you want to start creating your own certificates that it won't work. All I'm doing is keeping my ancient version of Outlook working with my email provider who is threatening ever newer TLS standards. Ben.
-
It lives... I didn't have to do anything special. This is using it on XP SP3 (with POSReady updates) unchanged as downloaded from https://github.com/josealf/stunnel-win32 Ben.
-
Thanks for the advice. I believe the latter versions of sTunnel support TLS 1.3 (I'm currently using v5.49 – the last official 32bit version). I was interested in upgrading for the extra TLS support but I didn't really hold out that much hope. It's kind of ironic that the very place where things like sTunnel are the most useful aren't supported. Ben.
-
Thanks for the hint! Out of interest, which browser did you use? I finally got to see the screen in your screenshot but only by using Firefox 116. Ben.
-
I was wondering if there was a more up to date version of sTunnel and came across this: https://github.com/josealf/stunnel-win32 ..which shows a recently publish version 5.70 but for the life of me I can't find anything to download. Does anyone see how to get: stunnel-win32-5.70-openssl-1.1.1u-installer.exe? Ben.
-
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
With the advent of BigInt support it is still not possible to log into a Fritz!Box, it still reports: TypeError: Promise.allSettled is not a function and fails to render the password field on the login screen. My current workaround is to use Mypal68 to talk to my Fritz!box when using XP. This suggests that there is some kind of edge case causing Promise.allSettled to fail but I'm all out of ideas as to how to debug it further. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
The only thing I'm aware of with Serpent is having to use: media.mediasource.mp4.enabled set to false otherwise the sound is distorted in a way that could be described as slow. In any case, setting this to false fixed the sound problem for me. This is a good description of the problem: https://askubuntu.com/questions/1200048/firefox-in-xubuntu-is-playing-mp2t-ts-webm-video-files-with-a-deep-slow (I don't use Linux) Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
That is exactly what I'm seeing too. Under FF52 I use dom.indexedDB.enabled set to false to block indexedDB – it breaks very, very few websites which suggests indexedDB is mostly used for gratuitous purposes – but that no longer works (because the problem is supposed to be 'fixed'). How do you do that? When I remember I simply manually delete everything in the Serpent profile's storage/default folder. Something automatic would be better! Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
A long time ago I used to routinely clear the Profiles' storage\default folder because it accumulated a lot of data – it's surprising how many web sites quietly use IndexedDB to persistently store who knows what in the profile. Later versions of Firefox handle these sneaky storage areas properly by tying there preservation or destruction to cookies. An example is https://web.lloydsdirect.co.uk (there's no need for an account, the IndexDB stuff gets created just by landing on that page). According to https://repo.palemoon.org/MoonchildProductions/UXP/issues/765 these changes were implemented some years ago but I'm seeing what looks like the old behaviour in Serpent. As far as I can see the storage\default folder is unconditionally accumulating data. Is anyone else seeing the same thing? Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Thanks for your reply. I'm not sure it's going to be easy for anyone to do anything unless they have access to a Fritz!Box. Unfortunately for me, most of the knowledge about these boxes seems to reside on German forums where, unsurprisingly, they speak in German. I'm still curious why the error I'm seeing (on the initial login screen) refers to Promise.allSettled and not to Promise.Bigint. Indeed, the faulting javascript, WebComponent.js, doesn't even reference Promise.Bigint. This is a shame as the Fritz!Box is an amazingly capable router/VOIP gateway ... I'm currently trying to set up VOIP. In the meantime I'll have to suffer the inconvenience of using my W7 box. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Thanks for your suggestions but I already have a solution, and that is to use the latest Firefox browser. However, I come here to understand how to extend the capability of browsers using XP – isn't that the point of this forum? I got past the screen in https://github.com/martok/palefill/issues/83 by blocking Fritz!Box's js/browser.js that puts up that useless message. In the console I can then see the Promise.allSettled error coming from components/WebComponent.js (but nothing about Bigint). Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
I'm having problems getting the Fritz!Box (Fritz!OS 7.50) login to work under Serpent v52.9.0 (2023-05-12) (32-bit), it's throwing this error: TypeError: Promise.allSettled is not a function According to: Implement Promise.allSettled() #1725 this was implemented a couple of years ago. Does anyone know what I am missing? Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Setting browser.tabs.tabClipWidth to zero is the answer. I don't normally open so many tabs that it makes a difference and only noticed because I was experimenting with the recent tab changes. Anyway, mystery solved. Thank you. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
@VistaLover thanks for the suggestions. I do use CTR, TMP, S4E, all the popular extensions that attempted to undo Mozilla's efforts to 'Chromeify' the Firefox UI. But this effect is odd because it only occurs once many tabs are open. It can't be just a CSS thing, so I don't know what it is. Yes, or right-click a tab and use Close Tab. Lots of ways to skin this cat. The thing about a UI is that there is often no 'better' way, just different ways, and most importantly, the way we get familiar using. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
Tabs are looking good and behaving again, although I have added this style: .tab-close-button.close-icon:not([pinned]) { display: -moz-box !important; } This makes the close icon appear on all tabs no matter how many tabs are showing. Without this, if I keep opening tabs, eventually the close buttons disappear on all but the active tab. I don't know what caused this behaviour to change or when it changed - it might even be something peculiar to my setup. I don't think it has been caused by the recent changes. Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
I think @basilisk-dev is now looking at it... https://forum.palemoon.org/viewtopic.php?f=61&t=29545&p=237431&hilit=tab+mix+plus#p237431 Ben. -
My Browser Builds (Part 4)
Ben Markson replied to roytam1's topic in Browsers working on Older NT-Family OSes
There's mention here: https://repo.palemoon.org/moonchildproductions/pale-moon/issues/1908 of a pref: browser.tabs.fadeLabels ...which supposedly turns off the new tab effects but I can't get it to do anything. The comment is a bit vague in that it doesn't say what type the pref is (true/false, 0/1, something else). I also can't see this pref being referenced in the code changes. Will this change have hit Pale Moon/Basilisk proper yet? Ben.