siria

LOL, couldn't help drafting such a disclaimer novel ;-P "For any help, support and discussion of roytam1 browsers, the best place is here: (link to NewMoon topic in msfn) Attention: Do not under any circumstances bother the PaleMoon team, especially Moonchild and Matt A. Tobin, with anything related to this forked browser, to anything XP-related, or mention any of roytam1 browser builds in the Pale Moon forum. They intentionally removed all XP-support for their products a long time ago, and do NOT tolerate any longer to be bothered by people using forked builds, especially not forks with restored XP support. Only discussion of their own original builds is allowed there. They have very STRICTLY declared this multiple times, in their own forum and also in the MSFN forum. So, please, abide by their request and go to MSFN only for support."

Seriously? That's dev speak, no normal user will have any clue what especially "upstream" means, and be confused yet more. And riddling what those mysterious other code sources or organizations might be.

@DrWho3000: you need to add the full domain to the override pref name, incl. ".com" or whatever a domain ends with, not just the first part. Subdomains in the beginning can be omitted though, e.g. "www." As others have written many times above, in clear examples. SITE USERAGENTS in general There are several possibilities where they can be defined (just like any other pref too): 1) A bunch of predefined site-UAs in original browser builds come inside omni.ja. If unzipped can be discovered hiding in a giant melting pot file with almost 2000 default prefs. File name goanna.js, or greprefs.js, all.js etc, depending on the browser version. Update (thanks Mathwiz in next post): ...or in newer browsers in file (browser installation folder)\browser\omni.ja\defaults\preferences\(browsername)-branding.js - or not at all in omni.ja anymore. Those default prefs can also be viewed and customized on about:config, unless overruled by 2) to 4) Omni.js is by far the most user-hostile place :( But at least it has the lowest priority. The prob is that just about no user is aware where those mysterious default site-UAs are coming from, and especially how to get rid of them when later becoming outdated. Would be much easier on users if fork-builders would define all their predefined site-UAs instead in a separate file, in place 2: 2) The best possibility to define user friendly default prefs: written in any file in folder ...firefox.../browser/defaults/preferences/.....js That place makes them "default" values too, and has priority over 1) In very old browsers the path is ...firefox.../defaults/pref/....js For easy maintenance users can simply create a new file there themselves, e.g. called myuseragents.js 3) the normal way for users: define personally customized site-UAs in the profile folder. By manually adding them in about:config, or by using an extension. Adding or changing them in about:config will automatically store the new values in file prefs.js in the profile folder. (except if now matching the default value again, then the browser will delete them from the profile, but that's another story) 4) very unusual but possible: users can store any prefs that are important to them in the profile in user.js. That will make them "startup prefs". Those can be changed during a session in about:config, but at next browser start the value from user.js will be imported again. 5) If an Addon is used for managing useragents, most of those seem to store settings instead in their own ways, in files in the profile, overruling Mozilla's pref system. (Tip for K-Meleon users: my macro useragents2018 manages only the 'normal' mozilla prefs, like the old traditional KM macro too, and leaves the handling itself to the engine) That depends. Have examined exactly this prob last year, and find it a huge prob that predefined ones are tossed nearly inaccessible for users inside omni.ja :( Yes, right-clicking in about:config and resetting a pref to default will delete all custom values. If a site UA has no default value at all, it will be deleted as expected, and afterwards the general override UA can kick in again. Catch: but if a site-UA does have a default value, defined in 1) or 2), resetting it in about:config will only reset it to that value again. The trick is: do not reset, but instead set it to a completely empty string. Now the general UA seems to work again.

Just quick links for roytam's retrozilla-versions, but that's a confusing mix of 2 authors and browsers with partly same version numbers for different builds. Just hope am getting that right: Retrozilla is actually the project of another dev, rn10950, hosted at github and here: https://msfn.org/board/topic/174987-retrozilla-an-updated-version-of-mozilla-for-windows-95-and-nt4-22-released/ It's an old Seamonkey (FF2) version, but with special updates, aimed for Win95/NT4 in modern web Roy RZ 2.1 = rzbrowser = fork of retrozilla browser Firefox2 + TLS updates http or https://o.rths.ml/gpc/files1.rt/rzbrowser-tls12-20180504.7z "About" says Firefox Community Edition 2.0.0.2, with a girl cartoon (Hmm, thinking about, neither the filename nor about page call this retrozilla...) Roy RZ 2.2 = rz-SUITE = an old Seamonkey with TLS1.2: http or https://o.rths.ml/gpc/files1.rt/rz-suite-v2.2-bin-20180708.7z "About" says Retrozilla 2.2, Seamonkey 1.1.19, Firefox 2.0.20, with a big red star But meanwhile exists a second retrozilla 2.2, by the original retrozilla dev rn10950 again. This Seamonkey includes his own updates plus Roy's pull requests from github (if I got that right, no clue of such stuff) Not retrozilla, but the same RZ topic contains somewhere also roytam's special Fx36 build, custom Firefox 3.6 with TLS1.2 http or https://o.rths.ml/gpc/files1.rt/fx36vc71-20171108.7z "About" says Firefox Community Edition 3.6.28, with a girl cartoon (same as RZ 2.1 above) (There are more special builds in that folder, see last link in previous post)

Thinking about, this topic would make a nice place for an overview of roytam's builds :-) There have already been some great postings across the forums, perhaps can be linked here. What I'm unsure about are especially the differences to the original builds. Just know that for some VERY old browsers (even FF2) his goal was to add modern TLS1.2, making half the web which is broken with "cipher errors" now readable again! Amazing how accessible the web can still be, if it weren't for such articial blocks. If the monopolies would only want to, instead of blocking without need and even deleting all older stuff. And so far his builds contain only engine updates, the GUI remains untouched. Nearly all of the engine updates are also collected commits from other devs and projects, like waterfox etc. Still the mass of browsers he manages to update regularly just blows my mind. My personal interest is mainly: ------------------- K-MELEON --------- KM-Goanna exists in 2 Roy-versions too: KG76+KG74 Attention Confusion: there are also much older K-Meleon GECKOS with the SAME version numbers, by KM-dev Dorian, called K-Meleon74 and 76 (engine 24esr+38esr) - Roy-KM-Goanna 76.1.1: palemoon27 engine plus selected updates, shell of Dorian's KM76RC2 from 2016 for XP SP3 or newer (SP2 with tricks, if a special OS dll copied over from SP3) http or https://o.rths.ml/kmeleon/ http://rtfreesoft.blogspot.com/search/label/kmeleon KM-Forum for latest 76.1.x http://kmeleonbrowser.org/forum/read.php?19,148500 MSFN-Forum mainly in NEWMOON topic, a bit in others too - Roy-KM-Goanna 74: final palemoon26 engine plus selected updates Shell of Naruman's KM1.8, a very customized old user-fork of Dorian's KM74 (incl. hardcoded ABP in browser/components, etc.) for Win2000 or newer (partly also usable on 98SE with latest KernelEx, very shaky yet, but best TLS1.2) http or https://o.rths.ml/gpc/files1.rt/KM74-g22-20180718.win2000.7z http://rtfreesoft.blogspot.com/search/label/kmeleon KM-Forum KM-Forum http://kmeleonbrowser.org/forum/read.php?19,146040 MSFN-Forum mixed in various topics ------------------- Guess there's also a pure Palemoon26 for Win2000 build. Most such exotic one-time builds can be found here (list needs JS+XHR enabled): https://o.rths.cf/gpc/files1.rt/home.html Or without JS but pure xml: https://o.rths.ml/gpc/files1.rt/index.php?sitemap Retrozilla is especially confusing with different authors/builds/same versions numbers, but deserves an own post too

No, you shall create those manually, brandnew. 1) c:/programs/firefox/browser/defaults/preferences/config-prefs.js content: the 2 prefs above 2) can't remember where, guess either in folder firefox or folder browser...

Just a general note regarding cipher quality in old browsers: on 98SE with kernelex my old main browser always was K-Meleon1.6 (engine FF3.5), which of course is the most horrible regarding ciphers. Those original old browsers now get kicked out by about half of all websites today, number still growing rapidly, and that means no access at all. It became absolutely disastrous only since 1-2 years, when google used its monopoly to force even the most harmless and public sites to switch to httpS only. And obviously most use only the most modern and strict TLS versions, providing no fallback for older browsers. Result is no access to important tools and resources on sourceforge anymore (except direct download links, if known), no info on dev-mozilla anymore, and MS killed github-access too after buying them, major disaster. So when a site now is blocked for FF3.5, the best solution used to be Opera12.02 (early TLS1.2) for many years. What I sadly learned far too late. Same for Firefox9. Both needing only basic kernelex for 98SE. If sites are blocked for Opera12.02 too, the next better browsers with modern TLS1.2 are now roytams old Firefox versions (retrozilla /FF2, Fx36, seamonkey). They even run on original old systems, without KernelEx. But if even those are blocked, which now slowly happens on a growing number of sites too, the only chance for 98SE I currently know of is roytams TLS-updated PaleMoon26 engine, which I'm using as KG74 (KMeleon-Goanna74 for Win2k) Definitely the best browser now for today's ciphers. The big prob is it's meant for Win2000, not 98SE, where it only works partially and often crashing, and also needs a very recent KernelEx version. But today this browser is often the last rescue when all else fails with broken ciphers, luckily. Just not usable as daily browser yet, crashing on me after a few pageloads, even without any JS. But I do feel it basically WORKS, there's not much missing yet! Just a little glitch somewhere... The engine itself seems to run fine, the prob is rather the browser shell or msvc-dll-stuff. Have zero dev skills, just suspect this because it happens mostly when right-clicking or opening some menus, and the titlebar can only show 1 character. Then again, am still on KernelEx v16, which is quite buggy itself. Meanwhile KernelEx v20 is out, probably better, just couldn't test it yet. ----- As for youtube, am mostly downloading them for viewing, with a userscript (gantt) and a macro in KM1.6 Until recently 3gp was among the choices too, great for small files with bearable image quality, but now killed by google too, arghh. Now only bigger mp4 files are available in desktop view. 3gp still exists on youtube's mobile view, but find no size choice there, only tiny "stamp" size. But unlike desktop view, this 3gp link on mobile view can be clicked to open in VLC and runs as stream. ----- Just sharing some experiences on 98se, by no means "the best ways" or anything.

It's not an "intruder", the profile is merely unknown. As they clearly say. It just means the server doesn't recognize that the current browser profile is still the same as at last login. Obviously that browser is better at reducing fingerprinting, has stronger privacy settings. Of course, no fun. A while back I actually had the same prob with amazon, with another browser (KM1.6 due to old OS), and drove me crazy too. Old cookies just ignored. Finally got it fixed somehow, it was some strange setting, but now don't remember anymore which one :( Possibly this one, although makes no sense: network.http.accept.default (am now using */*) Anyway, would compare your settings with your 'working' browsers. Can be anything, from different adblock stuff to cookie settings, storage permissions, indexeddb permission, each global or site specific, useragents, referer permissions etc. Check privacy and security addons. It's all about settings. By the way amazon is playing with their coding again, new quirks for ancient browsers (like my KM1.6) started 2-3 weeks ago: with old useragent string it now shows constantly bot-checks. Weirdly, the frequency seems to depend from time of day?! During daylight hardly ever, at night far worse. When toggling UA, claiming to be FF52, and reloading the page, all is instantly fine again - even without bothering to solve the bot task. But sadly cannot find any workaround for their second new quirk, showing on my old browser search results only as mobile view, 1 column, 320px wide :( Instead of the grid view for desktops. Perhaps "mobile" is their new default layout if the screen size is 'only' 1024px...

Perhaps worth a look, free editor with syntax highlighting etc.: 920 Text Editor https://android.izzysoft.de/repo/apk/com.jecelyin.editor.v2 https://play.google.com/store/apps/details?id=com.jecelyin.editor.v2 I see it's now 5MB, but when I installed mine four years ago it was just 1MB. That version runs quite well on my old 0.5gb phone.

(offtopic) I can't bear spying, but there seems no escape anymore, be it by Google/Facebook/Microsoft etc. or China - all the same, all stealing whatever they manage to grab. It's ironic, but I'm actually considering to buy a Xiaomi phone now: simply because their current Android One phones (Mi A1, A2...) seem the ONLY ones which even people like me have a chance to unlock without a giant struggle! Necessary for installing a custom ROM, which then hopefully has a lot less spying build in. All others make unlocking next to impossible (for me), or there are no custom ROMs, or phone features are missing, or crappy quality or wrong size or simply too expensive. This easy unlocking permission is extremely rare so am afraid only available for a short period.

Actually gets worse. The final blow for those who didn't need much JS-heavy sites and still hung on, is modern SSL/TLS. Since last year TLS1.2/1.3 got enforced on more and more big sites, meanwhile guess half the web already and still rapidly growing, thanks to Google boss. On Github, sourceforge-downloads, dev-mozilla, palemoon-forum, etc., everywhere. That's what really KILLS everything, no chance to even see some plain html anymore - nothing. Too bad. Need K-Meleon for customizing and tweaking stubborn sites readable again, but its last version working without huge OS probs or alpha-bugs is KM1.6 (generation Firefox 3.5) Now luckily there are a few TLS-enhanced browsers like Opera12, and especially Roy's updated builds like Retrozilla, which demonstrate that it would still be possible to browse 90% of todays web 'well enough' even with FF2 engine (plus some tweaking), which is amazing! If only many pages weren't so cluttered with big unremovable boxes hiding the text content behind them, everywhere big menu blocks and fixed title blocks and footers and teaser blocks etc. Without KM have no chance to chase those away, or insert little js-snippets and custom css and calculate redirects etc.

May well be, just as another possibility, a pref not visible in about:config doesn't necessarily mean its default value isn't active anyway - just hidden from users yet deeper. Like all the "permissions.default.xxxx" prefs except "image" (others are subdocument, script, xmlhttprequest, stylesheet, media, object, etc. Values if visible are 1/2/3, but with "invisible" working like 1 too) About a mysterious hidden place with lingering cached effects from addons, some candidates: in the far past it was definitely xul.mfl in profiles (perhaps along with another *.mfl), and later the startupCache folder in profiles, and thought I had once discovered similar cached stuff in "Local Settings" folder but then couldn't reproduce it anymore, and also not sure anymore which version and how. Perhaps because my profiles are usually portable.

Yes absolutely (just rename inside user_pref to pref) but important to understand: - prefs.js (and user.js) are USER values, stuff inside always has priority over default values - but prefs.js only contains prefs that are NOT default (or would contain 2000 lines) - after changing default values by adding this file, if prefs.js did already contain the same setting (user set), it will now be considered 'default' and the line vanish inside prefs.js Very easy and safe to test: if you make a backup of prefs.js Then just add your new defaults file in preferences folder and restart. For testing, it could contain anything, your real contents, and/or for example: pref("permissions.default.subdocument", 1); pref("permissions.default.subdocument__INFO", "INFO: 1=load all iframes / 2=block iframes / 3=same domain"); (that pref and siblings are global default values, working natively, but probably conflicting with certain addons for the same stuff) PS: I'm not a programmer either, far from it (except KM macros), just also a user who loves customizing ;-) And find this basic pref knowledge far too unknown, which is a major pity. Heard of that stuff myself far too late, only after years of having missed especially startup prefs - while it would have been so easy since decades already. Just didn't know until some day stumbling about it accidentally, sigh. And not sure if you're interested, perhaps for comparing notes, but actually such projects (optimized pref collections) are posted from time to time in various forums, or blogs like ghacks.Everyone has a slightly different approach, most focussing on speed, or privacy, and guess yours adds a new twist - old machines.

Just a little tip for a better installation way, because I feel you're not very aware how prefs.js works. For example this file only contains user prefs if they are different from default. All other lines (matching default) are automatically deleted by the engine. instead of profile/prefs.js, it has advantages to put the file instead in browser/defaults/preferences/looking4awayout.js or UOCpatch.js The syntax is the same except it's only "pref()" not "user_pref()", example pref("zoom.defaultPercent", 100); The advantages are - you can keep those settings together in one place, the lines inside never vanish, and you can easily look it up later (Ah that was a uoc-pref...) - all settings in the defaults folder are automatically "default" prefs (of course, users should remember this!) - you can write comments inside, because the browser only reads default files, never writes inside - default also means, unlike prefs.js, the lines don't vanish in nirwana as soon as you toggle something to the default value. A toggled value, if non-default, just becomes a "user setting", which means it's now written into prefs.js. To restore original UOC, you can just right-click prefs in about:config and reset them to 'default' again (as defined in your file). Resetting to default will also delete the line in prefs.js Another possibility: adding the file contents inside profile/user.js Then it doesn't mess with default settings for all profiles, only affects the current profile. An important difference: profile.../user.js contains startup prefs. Users can toggle them during a session, by menu or in about:config etc, but at next browser start user.js values will get imported into prefs.js and become active again. A little trick about prefs.js: if entries were manually changed with a text editor, and there are duplicate lines, the browser will clean it up automatically at next startup. The later lines 'win'. That means: instead of replacing prefs.js completely, you can just paste a whole block of new prefs at the end of it (while the browser is closed!) Then all previous user prefs, if they were NOT duplicated in the new block, are not lost but will survive. PS: If you try K-Meleon-Goanna, highly recommand to download macro "useragents2018.kmm", instead of installing any conflicting header/UA addons/extensions. Unlike them, this macro only toggles prefs, nothing else. The UA spoofing itself is done by the native engine.

If all else fails or is forbidden on certain domains, spoofing useragents in Firefox52 is still easy without any extension: general.useragent.override = Mozilla.... Raw basic method. That UA-string simply gets sent to all domains, not just single ones. Works since ancient times and in FF52 too without any additional initialization, unlike site-overrides. The only catch is that all those header addons must be disabled to avoid hijacking. As usual, after addon install changes it often helps to delete their startupCache folder in the profile. (Being stuck on a much older browser myself, struggling with broken websites all the time, it's often even necessary to kill the stylesheets after loading. For example here too, to get the Edit link ;-) Also a very powerful method to get some hidden or broken stuff partly accessible again)

Yeah, it's definitely fully intentional. Microsoft is known for trying all sorts of tricks to enforce the newest browser and OS versions (with maximized spying tools) on all people, if any possible. So the first thing they did after getting their hands on github was to enforce a newer TLS version, even for simply *reading* public pages. In one scoop kicking out all users of really old browsers knowing max TLS1.0 (perhaps 1.1 too, don't know), and even worse, also making it impossible for them to download anything from github anymore. Countless useful apps developed for OLD browsers now inaccessible, and updates to those, and even Linux builds. And shortly after people here reported that Microsoft now also blocked access to a few of the advanced github-features for members, by enforcing really latest greatest javascript and the most modern browsers on them, if I got that right. That fits also perfectly to this discovery that Microsoft-Github now only allows full access for the current Firefox ESR 60 version too, by intentionally using most modern JS code, and even the previous FF-esr is already ignored! They clearly do this by design. Trying to enforce as much as they can their latest systems and browsers with maximized spying build-in and minimized customizing possibilities for users. Like others too of course (e.g. Mozilla deleting ALL xul-addons which makes no other sense at all, instead of simply keeping a frozen archive, and Opera deleted all Presto-addons long since too) Regarding really old browsers again, luckily one of them could already handle some TLS1.2 and html5: Opera Presto 12.02. The prob is, that's not widely known, sadly I learned this only last year. And meanwhile, what hardly anyone knows, there's also retrozilla (FF2) and a few old gecko builds with added TLS1.2 by roytam1, working miracles in modern web, even on 98se with kernelex: you wouldn't believe how many sites become at least readable again simply by being allowed access. Without frills and only limited functions of course, but blocking css sometimes works the next miracle, revealing input boxes, text content, many images, making some broken links clickable. BTW also making some too heavy css-overloaded pages scrollable again, fast. For miracle step #3 fiddling with userstyles+scripts often helps (for this forum too) All that stuff doesn't help always, but well enough in most cases. Point being: Yes they are fully intentionally trying to kick out users of even slightly older browsers versions, for their own hidden agendas, without real need.

That makes perfect sense: The coders use some if-rules for certain defined alternative browsers, and all the "unknown" rest gets the "default" value - for current modern code. In my old KM-version I've also noticed that spoofing as IE7 gives very often better results as spoofing as FF3.5, matching the real engine. For example on amazon spoofing IE7 makes the article's main image show up, otherwise it's invisible. My theory is that quite some website-devs still just keep inherited ancient if-rules for IE7 from old times, when it was the most important browser, by simply not touching those lines, and despite some rendering quirks the result in old gecko browsers is now a lot better as the modern "default" code for newer Firefoxes. Who would have thought that one day it would be a good thing that old IE wasn't strictly standard :-) For old Firefox-versions such if-rules were not necessary at the time, so today they land in the big pot who gets just modern default code.

How to initialize site-specific useragents again http://forums.mozillazine.org/viewtopic.php?f=38&t=3031074 yes basically all it takes to initialize the sleeping site-UAs again is running such a simple line at startup: Components.utils.import('resource://gre/modules/UserAgentOverrides.jsm').UserAgentOverrides.init(); This works after input in error console after every startup. But I'd love to get just this line as a tiny, independant xpi-addon, which users of *old* Firefox and K-Meleon74-76 versions could use, and remains visible in about:addons. Ideally with on/off switch during session by observing a pref. Just how? In newer KM the initialization was already fixed by roytam1, in all goanna builds. But would be nice to also have a little fix by addon for users of older gecko 24-38 builds, without needing to update the whole browser, which can be a real pain in K-Meleon if heavily customized. This little init-addon would need no GUI, just this init-line. Users can either edit site-prefs directly in about:config or use already existing GUI-addons, and in K-Meleon use the menu created by macro useragents2018. Already tried dabbling with it myself last year, using blind trial&error, but this must be possible a lot easier and better: http://kmeleonbrowser.org/forum/read.php?2,146871 @sdfox7: Tried also "general.oscpu.override" = Windows NT 6.1 ? Perhaps they do feature sniffing, but there's also a whole bunch more override prefs available, for all sorts of details. Saw them listed somewhere... One of those sites like you linked, and also in an old UA addon, which can spoof those as well (would have to look it up again, perhaps pederick's?)

At the moment instagram still kindly allows all users to see their videos, even on Win98se. The catch is just that you cannot watch them directly inside a browser, but first have to download the mp4 files, from the direct link in the html source (metatag "og:video"), and then watch them with a local player. For me that's fine enough, considering it's the only chance - hundred times better than no access at all. Glad they still allow that, but they can stop it any day of course, like most other sites already did. By the way there are tons of interesting "og:..." tags in the sources, also direct links to the big main image etc., very useful for users on old systems. And not only contained by instagram and facebook but also a few other services. For frequent use the access can be made easier with a little javascript, like (function() {var vidurl, metaTags = document.getElementsByTagName('meta'); for (var i=0; i < metaTags.length; i++) { if (metaTags.getAttribute('property') == 'og:video') vidurl=metaTags.content; } return vidurl;})()

Regarding new names, haven't studied the old discussions yet, but would find it helpful to have some sort of recognizable family name. For XP-builds perhaps just XMoon, XFox... or RoyMoon - lol

Not sure if that applies in your case, just in general: most people are not aware that the "general" pref is overriden by site-specific ones. And especially not that a whole bunch of hardcoded domain spoofs come predefined already out-of-box For example roytams KG76.1 contains a spoof for amazon.com, sending FF45. The whole list can be found on about:config, filtering "useragent" For real general spoofing disable this one temporarily: general.useragent.site_specific_overrides (=>false) And also check the other "general.useragent" prefs, if existent, for sub-elements. Ah right and almost forgot trap-4: if useragents work like other site-specific stuff in geckos (for images, videos etc. in permissions.sqlite), then those domain-spoofs don't work top-down (only domain in urlbar matters) but bottom-up (source domain of single elements matter). That means even when surfing on amazon.co.uk, the spoof for amazon.com still affects those page elements that all country-amazons are loading from ".com" Or whatever other domains they are loading stuff from, for scripts or general resources etc., usually todays major sites use a whole variety, e.g. nearly all including stuff from googlexxx

No it often means nothing if something works in one browser and in another not, as so many people seem to assume. Sometimes yes, but very often not. Not everything depends from the system only, browsers can also bring some own capabilities too. Especially that encryption/SSL/TLS stuff is contained completely independent inside all Firefoxes and the forks, as I've been told here. Unlike other browsers which fully depend on system capabilities. That's the reason why Firefox forks can still be updated to use TLS1.2 on systems even older as Win98, and other browsers are stuck with what the old system can do. Yet lots of things which work fine can still fail if they simply aren't enabled in the settings. Settings are an incredible jungle in FF/gecko, there are over 3000 prefs, and some most important ones not even showing up in aboutconfig, and many cross-influencing each other. Clean installations also are no guarantee that everything is fully enabled out-of-box, as so many people assume. For example some of roytam's browsers for very old systems, like FF2, are now equipped with TLS1.2, yet out-of-box still have the ancient default setting of max TLS 1.0. Or stuff like hardware-acceleration, those settings are infamous to be helpful on one system yet disastrous on another, so intentionally not always enabled by default. At any rate, if the map first shows up fine and then vanishes again, the browser CAN do it, and it really doesn't matter what Chrome or the system can. The solution or the workaround to persuade the server to send the tiles must be somewhere in the settings....

Just a side note, in my old browser and on OSM that happens when the scripts switch to a higher encryption, which it can't support yet. First looks fine, then BANG - server kills the tiles. Maybe check current TLS1.2 settings? "security" prefs in aboutconfig. Hm, and remember some pref about security hardening... Another thought: Fake the useragent. Some major sites today just kick out XP-users (if UA contains "Windows NT5.x") Or perhaps the browser already has a hardcoded fake UA for the tile server...

If roytam writes to add this file IF it complains, then I would surely only copy it IF it complains, otherwise makes no sense. And hey, never change a running system ;-) That file directory on orz looks exactly like the one on o.rths.cf, also needs JS+Ajax, probably same server with different redirect domains. Anyway, good to have a fallback. What I forgot to mention, but jumper above wrote it already: 1) roytam's files are available with or without httpS-encryption! :) If not possible to download a file from httpS URL, simply change it in urlbar manually to http 2) And he got a new domain a while back, all URLs changed from "o.rthost.cf" to "o.rths.cf" If something doesn't download, that's also important to remember. Just changing domain manually too. Guess it's probably better to post the direct links too... All builds by roytam1, with added TLS1.2 if I got it right. That means nearly all of those pesky errors are fixed with his builds: "Error code: ssl_error_no_cypher_overlap" HA!! :P Firefox-3.6.28 for NT4 (may somewhat working in NT 3.51): http://o.rths.cf/gpc/files1.rt/fx36vc71-20171108.7z (confirm that it runs fine in Windows98 fine too! (tested in 98SE) Firefox2 / rzbrowser / Retrozilla-2.1 with TLS 1.2 support: http://o.rths.cf/gpc/files1.rt/rzbrowser-tls12-20180504.7z (much older engine as in FF3.6) Seamonkey / rz-suite / Retrozilla-2.2 with TLS 1.2 support: http://o.rths.cf/gpc/files1.rt/rz-suite-v2.2-bin-20180708.7z (has a mail client etc. included)

Yeah roytam's lesser known builds are kinda hard to find, but finally managed to find them again - Here! :) LINK 1: http://rtfreesoft.blogspot.com/2018/06/other-browser-binaries.html At bottom of post: Firefox3.6 + TLS1.2 (fx36) Firefox2 + TLS1.2 (rzbrowser / retrozilla2.1) And better yet: the archive link! Can't believe how long it took me to notice, that this folder isn't as locked away as thought for so long, the trick is just: it needs AJAX for the content to show up! LINK 2: http://o.rths.cf/gpc/files1.rt/home.html There's also Seamonkey + TLS1.2 (rz-suite / retrozilla2.2)