siria
Content Type
Profiles
Forums
Events
Posts posted by siria
-
-
(_HOWTO_ modern ciphers for old browsers for Windows98, Win98, Win9X, ME, forks, TLS1.2, ...)
ZortMcGort11 said:Are there any Windows9x browsers that load Wikipedia anymore? Are you guys still able to load wikipedia.
YES - again
It was starting to become a real killer prob 1-3 years ago, half the web broken, even with basic KernelEx, but thankfully roytam1 appeared, and has posted a bunch of updated OLD browser forks! Adding modern TLS1.2 CIPHERS, and he also backported a few browsers to work on older systems again. That stuff is just shattered around here and there, a bit everywhere...
For VANILLA Win98 there's mainly Retrozilla (project of dev rn10950), in various flavors, and roytam's recent fork of K-Meleon1.5.4+TLS1.2, with modern ciphers added.
There exists also a roytam-fork of Firefox2, which is also called Retrozilla, with now an rzbrowser.exe inside. Its older version still had a firefox.exe inside. Yes it's a bit confusing...
RETROZILLA (ancient Seamonkey, era Firefox2, but with modern CIPHERS):
guess this exists now in 3-4 flavors, but am rather confused, since initially, when I found and used it for awhile as fallback browser, had no clue yet there exist 2 RZ-builders with completely independant forks, using even same version numbers, and different engines inside! Had completely lost overview in that chaos and given up when not needing RZ myself anymore, but finally did some more research now:
The official Retrozilla is a fork of an ancient Seamonkey version, by rn10950, hosted at github, and he started the msfn topic here in 2016:
https://msfn.org/board/topic/174987-retrozilla-an-updated-version-of-mozilla-for-windows-95-and-nt4-22-released/
Not long ago he posted an update to version 2.2, containing modern CIPHERS and other stuff.
But independant of this version, roytam had already made some own RZ-forks too, before and after the official RZ update, and partly used other engines inside too, and also called them version 2.2.
Guess there exist now
1) RZ-ori-SeamonkeySuite, and
2) RT-RZ-SeamonkeySuite, and
3) an RT-"RZbrowser" with Firefox2 engine inside, which is even more confusing, because:
I happened to stumble upon this "RZbrowser" first, so used it, unaware of any other builds/forks/variations, and at the time it was the older version yet, v2.1 from 20180504. Was completely confused that it didn't identify itself as RZ, but instead as "Firefox2 Community Edition", and was started by a firefox.exe, and it also stored its Profiles in AppData/Firefox/. When I finally got it that this really is FF2, roytam posted an update to "Retrozilla v2.2". Now I was AGAIN completely confused: this update looked completely different as my "previous version 2.1"! And it was started with a "retrozilla.exe" and stored profiles in appdata retrozilla folder. Finally learned that there are two branches: "rz-suite", based on Seamonkey, and "rzbrowser", based on Firefox2. Awhile later, an update of this Firefox version followed: rzbrowser-tls12-20200127.7z. And AGAIN completely confusing: now this update identified itself as RZ too, in the startup page! And it gets now started with "rzbrowser.exe". So, no Firefox inside anymore? Then again, the exe properties still claimed to be FF2. And the toolbars still looked like Firefox too. After starting noticed the next surprise, it created completely new profiles in a completely new path: "AppData/Retrozilla/BROWSER/Profiles" (note that RZ-Seamonkey SUITE uses "AppData/Retrozilla/Profiles", and "retrozilla.exe") Uff, what a chaos... But what I mainly wanted to know: can this new version still run normal FIREFOX addons? And luckily, after a quick test: Yes! My test addon Read Easily installed quite normal. So it really is still a Firefox towards addons. Just was confused again for a moment that the new button wasn't visible, until realizing it must first be unhidden, by right-click on a toolbar and "customize". There it was hiding! Ouf....
So, this was the older and "pure" RZbrowser-Firefox2 version I used myself a few months as fallback, simply because it was the first RZ I ever saw:
https://msfn.org/board/topic/174987-retrozilla-an-updated-version-of-mozilla-for-windows-95-and-nt4-22-released/?do=findComment&comment=1150700
HTTP or https://o.rths.ml/gpc/files1.rt/rzbrowser-tls12-20180504.7z
And this is the newer RZbrowser-Firefox2-wrappedRetrozilla Mixup:
http://rtfreesoft.blogspot.com/2020/01/other-browser-binaries-20190131.html
http or https://o.rths.ml/gpc/files1.rt/rzbrowser-tls12-20200127.7z
And here is the original "complete" Retrozilla version, based on Seamonkey, containing a retrozilla.exe, by dev rn10950, but downloads hosted at Github (=now BLOCKED for old native Win98 browsers, thanks Microsoft!)
https://msfn.org/board/topic/174987-retrozilla-an-updated-version-of-mozilla-for-windows-95-and-nt4-22-released/
https://github.com/rn10950/RetroZilla/releases/latest
Perhaps direct links are better for future...
retrozilla-2.2.en-US.win32.installer.exe 12.5 MB
retrozilla-2.2.en-US.win32.zip 11.4 MB
Source code (zip)
And roytams forked version of "complete" Retrozilla SM suite:
http://rtfreesoft.blogspot.com/2020/01/other-browser-binaries-20190131.html
HTTP or https://o.rths.ml/gpc/files1.rt/retrozilla-suite-tls12-20200131.7z
The main differences between RZ-Seamonkey and (RZ???) Firefox2, as vaguely as I understand them, is that SM is a suite and comes with an email tool and chat etc build in, while Firefox2 is "only" a browser, but for it are still many more addons available, in wayback machine. Incl. crucial ones like CustomButtons and "Read Easily" (toggle styles off permanently, with button). While addons for SM are still found on Mozilla's official AMO site or guess an official split of it, among Thunderbird addons.
Now, not sure which Retrozilla version you tried, if you found it so bad, perhaps on old one.
Of course, all engines inside all versions ARE stone age, around Firefox2 era! Just updated with modern CIPHERS to restore access to HTTPS sites. And the original version probably yet more updates.
Then again, you mentioned too many CRASHES, but crashing is a whole different prob in itself, sadly more and more frequent now. KM is heavily affected too, old versions and newer ones. I hear even medium aged browsers like current KMG76.3 (based on PaleMoon26+Updates) and running on modern systems have started crashing here and there
roytam's K-Meleon154+TLS1.2 ciphers is here
HTTP or https://o.rths.ml/gpc/files1.rt/K-Meleon1.5.4en-US.tls12.7z
(3 files missing: profile.ini for portable mode, msvcp71.dll+msvcr71.dll, if not in system folder already. See below)
KernelEx NOT needed on Win98, but if you run it in a kernelex system anyway, set the compat to "disabled" to avoid a bug.
Anyway, if KernelEx is installed, choices get much better:
For example roytam's version of Firefox3.6+TLS1.2 ciphers, build for NT-systems.
Perhaps runs on vanilla ME too? But have heard not on all, so suppose Win98 still needs KernelEx?
HTTP or https://o.rths.ml/gpc/files1.rt/fx36vc71-20171108_newnss.7z
Incidentally this also contains the 2 mscv files missing in KM1.5.4 above.
For portable use it can be started with the contained bat file.
And for KM-fans.
A tiny step in numbers, but huge step in functions: K-Meleon 1.6+TLS1.2
You don't happen to still have an old community updated version of KM1.6beta2.6 lying around? Downloads of the time are long since broken too. Only the outdated official 1.6beta2 on sourceforge still available in repo, but lots of fixes and updates missing:
K-Meleon 1.6.0 Beta2 (Dec.2010) Win98 needs KernelEx452 with compat=XP
HTTP-SF or https://master.dl.sourceforge.net/project/kmeleon/k-meleon-dev/1.6.0Beta/K-Meleon1.6.0_Beta2_en-US.exe
HTTP or https://master.dl.sourceforge.net/project/kmeleon/k-meleon-dev/1.6.0Beta/K-Meleon1.6.0_Beta2_en-US.7z
Important Update packages:
(but sadly only the most crucial updated files. There were more in old times, but download long since broken):
2012 +Beta2-3: http://kmeleonbrowser.org/files/KMeleon1.6.7z (by Dorian)
(2012 Gecko updates by forum members, beta2.6 by JamesD, download broken)
2020 +TLS1.2 ciphers: HTTP or https://o.rths.ml/gpc/files1.rt/ns9-nss-update.7z (by roytam1)
Attention: For KM REMOVE sqlite3.dll from nss-7z, KM must keep its original version
KM1.6+TLS1.2 topic
http://kmeleonbrowser.org/forum/read.php?2,151536
And for CometBird9+11, Firefox10, Netscape9:
Some little NSS-update packages by roytam1, adding modern TLS1.2 CIPHERS:
https://msfn.org/board/topic/178283-how-you-really-browse-the-web-on-98me-in-2019/?do=findComment&comment=1178208
HTTP or https://o.rths.ml/gpc/files1.rt/CometBird-9.0.1-new-nss.7z
HTTP or https://o.rths.ml/gpc/files1.rt/CometBird_11.0-new-nss.7z
HTTP or https://o.rths.ml/gpc/files1.rt/Firefox10-new-nss.7z
HTTP or https://o.rths.ml/gpc/files1.rt/ns9-nss-update.7z
If anyone has Win98+modern KernelEx2016:
roytam also forked a special K-Meleon74 build (based on an old KM74-fork by Naruman), backported it for vanilla Win2000 (=Kex compat) and added modern TLS1.2 ciphers.
In Win98 this one is far too buggy+crashy for real use, but sometimes the only chance.
Perhaps it runs better in ME?
He also fixed 2 old KM74-bugs, native site-useragents and macro stuff (injectJS-return).
HTTP or https://o.rths.ml/gpc/files1.rt/KM74-g22-20180718.win2000.7z
(files inside from 2020-01-09. If only minor (?) updates like a few more ciphers, RT often insists to keep old date in name, sigh)
With VERY modern KernelEx2019 (and perhaps ME, not Win98? not sure) someone even reported that RT-KM76g starts in his system, but also rather buggy of course... Since 2-3 years roytam1 updates the engine inside his Goanna3-fork almost weekly. For example, at the moment the last version is here:
http or https://o.rths.ml/kmeleon/KM76.3.1-Goanna-20200926.7z
ZortMcGort11 said:I'm using vanilla KernelEx, last version posted on Xeno's old website.... 4.5.2. Is there a simple download package that update from that?
Lots of packages, but very far from "simple". And when I finally dared to try updating last year, hoping to get RT-KM74 running, it nearly ruined my system. For an expert no prob of course, but for me, having no clue of system stuff, so decided to leave it at a semi-updated state and not daring to touch it anymore since. Killer prob is, my 2016 version is also rather buggy, not possible anymore to disable Kex completely on an app, killing some old native 98 apps
By the way, when using a BRANDNEW profile and getting this ERROR on all HTTPS sites:
"Could not initialize the browser's security component"
that's also a KernelEx bug, but luckily can be worked around. HOWTO:
NEW profiles must first create 3 personal key files for HTTPS (key3.db, cert8.db, secmod.db)
- either copy over those 3 files from an older, existing profile of a similar Mozilla browser
- or produce new files with a native Win98 browser, then copy those 3 new files over into your younger browser
(from Firefox2, K-Meleon1.5.4, Retrozilla? etc, into Firefox3.5/3.6/9, K-Meleon1.6/7/74 etc.)
For example, if using for this key-creation roytam's updated fork of K-Meleon1.5.4, in Win98 with KernelEx:
right-click on "k-meleonW9x.exe", open file properties, set KernelEx compat to "disabled", start this exe, and visit a httpS website, like msfn. If it works now, the browser has created those files in current profile folder. From there you can copy them into other browser profiles, of younger engines.
General info for vanilla Win98 users:
If UNZIPPING of roytam1 browsers doesn't work, update to 7-ZIP 9.20, last Win98-version:
HTTP-SF or https://master.dl.sourceforge.net/project/sevenzip/7-Zip/9.20/7z920.exe
or http://sourceforge.mirrorservice.org/s/se/sevenzip/7-Zip/9.20/7z920.exe
For running K-Meleon1.6, 1.7, Firefox3.x etc. on Win98, you need at least good old basic
KERNELEX 4.5.2 for Windows 98:
HTTP-SF or https://master.dl.sourceforge.net/project/kernelex/KernelEx/4.5.2/README.txt
http or https://master.dl.sourceforge.net/project/kernelex/KernelEx/4.5.2/KernelEx-4.5.2.exe
Ouf... what a novel, sigh.
.
1 -
ZortMcGort11 said:I'm using K-Meleon as well, version 1.7 and no javascript as well. Sometimes I'll use Firefox 3.6.28, but K-Meleon allows more tweaking and adjustments to the user interface, and the bookmarking works better. So I'm pretty happy with it. Opera 12 is useless now, it breaks down on almost every site I try, with all the certificate error messages and everything
Wait a sec.... just surprised... you're really getting MORE sites readable with KM1.7alpha as with Opera12.02??
And what about wikipedia? github? sourceforge downloads etc...? (KM-Forum?)
In any case doesn't sound as if you're aware of roytam1 cipher-updated forks of old browsers, with TLS1.2
And which KernelEx version are you using? Basic 4.5.2 or newer?
By the way this forum has quite a bunch of topics for Win9x, although most people are using it just for retro-fun, in a virtual machine or dualboot system.
(posted with KM1.6beta2.6+TLS1.2 by RT)
.
0 -
WEIRD, this black screen mystery!
Sorry have no clue of hardware/driver stuff, just wondering which versions of NewMoon/K-Meleon you use?
On Win2000 perhaps NM26 + KG74?
On XP the same...? Or NM27+KG76?
Curious what will turn out as reason, hope not malware. Why can it break so suddenly, in *several* browsers at once? Very strange. Even if it may be fixable now by modifying individual browser prefs, but the root cause remains dark. Perhaps a sudden driver hickup, but would that happen at same time as copying over browser builds?
Your best chance may be trying to remember as much as possible what you did exactly.
Cixert said:Today I put some versions of K-Meleon and New Moon configured in another computer and that is when it has stopped working.
I've tried your original K-Meleon and New Moon and now they don't work either
Interesting.... could it be that just the profiles are messed now with specific hardware settings, and you copied those over? Although by default KM has portable profiles inside program folder (unless you deleted file "profile.ini" or using commandline etc), and would assume you tried fresh profiles already. But just in case, perhaps test again anyway:
KM: Edit > Manage Profiles > New
Check "Ask at Startup" and restart browser.
Can you see that, or are KM menus black too?
In that case find file "profileS.ini" (with S!) and edit it with notepad:
StartWithLastProfile=1 / Change to 0
Or in NewMoon, trying Safe Mode....
Or use commandline to get ProfileManager at startup: ...\palemoon.exe --ProfileManager
Or simply make a backup of complete Profiles folder, then delete original and restart...
Cixert said:On Windows XP I do not have black screen problems, but it is also curious that in all browsers Google asks me to accept their privacy conditions every time I start them.
Cixert said:The problem on Windows XP has been fixed.
It was because I had set to delete cookies when the browser is closed.
I was confused because every time I opened a new tab with Google, the question would reappear, after I accepted on other tabs.
Now I have set keep cookies until they expire.
That's quite confusing too....
I understand that cookie pref makes sense for your initial post: That browsers ask everytime after you start them.
But riddling about your newest post:
Browsers also forget cookies when just opening a second TAB, after having accepted cookies already in another tab? And do they react differently if those previous tabs are still open or not?
0 -
Dave-H said:I don't actually mind the new design too much, I've seen far worse, but I do want the videos to work!
Just curious, regarding possible workarounds (can't test myself, old browser): does this new layout still contain the direct video links in html source code? Especially easy in meta tag "og:video". And futher down are SD and HD variations.
In whichever version they send me, those links still work for downloading. If giving a timestamp error, must just replace "&" with "&". Suppose a little userscript can also wrap those links into a simple html5 video tag
.
1 -
reboot12 said:Please do not write in this topic about browsers or plugins other than the latest versions of Firefox and Chrome for WinXP !!!
Well, as RainyShadow already mentioned, after 11 pages it's a bit late to suddenly convert a general topic into a very restricted one:
From "old browsers"
to:
- only Firefox and Chrome
- only on XP
- AND only their LATEST version for XP
reboot12 said:I should have titled as:
YT may not work on old browsers Firefox & Chrome anymore, starting March 2020
that's neglecting "XP" again, and "latest version"....
.
2 -
Wunderbar98 said:Logged in to HTML Gmail, seems to work too
Working gmail homepage on DOS, with a DOS browser?? Absolutely incredible!
Oh well, am probably not sys-savvy enough to even get a network connection, and should spend my time with UPdate stuff not downgrades, but even sitting on the sidelines it's fascinating just to know that's possible at all!
A few screenshots would be nice.
Hint:
the current title is completely misleading. As is the name of the browser itself of course. "Links" - what a non-name, sigh...
At any rate everyone who'll see this topic in the list and reading "Links for DOS Community Edition" is made to think this topic were for "Hyperlinks to some forked DOS-system" Title should somehow indicate this is actually about a DOS-browser with TLS1.2
0 -
RainyShadow said:I'm wondering whether to update since S52 and NM27 are my main browsers...
No need to hurry... KM-Goanna76 shares the same engine with NM27, and this "BigChange" build (which should IMO get a little version-number update) seems to have some severe new bugs:
http://kmeleonbrowser.org/forum/read.php?19,148500,153121#msg-153121
.
0 -
roytam1 said:Notice: since upstream merged a big branch, it will take more time than usual to get changes landing in my tree. as a result today's builds may or may not be done.
for v27(g3) tree, since upstream broke the build, g3 based builds are not available this week.
for boc, since upstream vanished its repo, I'm now thinking of building icedove-uxp instead.
Whoa, sounds like not just 1, but 3 major upstream disasters. Don't get why MT is out for revenge again without any obvious trigger so suddenly - just because, as he mentioned, he doesn't like the new branding "BNavigator", saying it lacks imagination?!
And he even broke NM27 (=KG76) too? Frankly I'd have more trust in new builds that just have "about:permissions" fixed, before merging tons more of (possibly questionable?) upstream updates
.
1 -
VistaLover said:
He pretty much (as expected) sides by M.A.T. because he was undeservedly banned from MSFN without notice, as being "a long-standing member with a good track record"
Undeservedly - har har... King of flaming rather. Long-standing member? May be.
But the irony is the "track record" is true acc. statistics:
who else here has a likes/postings ratio of even close to 1.5?? (115 Likes / 75 posts)
Which also gives him an "excellent" reputation on the profile. Everyone who hasn't actually read his old postings WILL be convinced he's one of the most liked members here, and not believe anyone who dares to claim otherwise. Every time I saw this likes-ratio I couldn't help but being reminded of Einstein too, this is SO much how today's society works. Since no dis-likes are allowed of course, that would hurt the feelings of such poor, highly sensible attackers too much. Who surely have only turned evil because they had a very rough childhood etc., or people confuse aggressive narcissists and dangerous psychos with mostly nice Aspergers who do NOT enjoy harming others, etc., but I digress ;P
Anyway, bans are rarely forever, and he's an important coder, am sure he'll be back again some day. Provided he wants to. At least he's very frank about his mindset. (Contrary to most other similar people, passing as Saints doing just incredibly "dumb" things unintentionally. All the time.)
The good thing for me personally, have learned from this addons-issue two technical tricks:
1) those pesky CLOUDFLARE errors 502 etc. do NOT necessarily mean any REAL server probs!!
It's possible, but just as well they can be created intentionally by website authors, even simply based on useragents or other settings. Actually this was the second such case in a week.
2) Important page content or even complete pages may also be hidden inside <template> tags, just like the popular SVG-wrapping trick. Users getting to see only an empty page, or mostly empty. Funny thing is, my old browser doesn't even know this tag yet, so renders it visible anyway. Normally. But awhile back had the brilliant idea to add a few "modern" tags to its global html.css, such as <mark> (for yellow background), but also <template> and <datalist>. The last two are set to "display:none" in modern browsers, so after a bit hesitation I just copied that too. Oops. Now old JS-engine didn't make the addons-page appear either. Still empty.
=> Now my user css file contains this:
template {display: block !important;}
HA: page appeared (yesterday), no javascript needed
(if users do have javascript enabled, on a younger browser, I suppose they may get double-content this way, but no idea)
-----
General trick:
Anyone else who must browse without JS, and running into empty pages all the time? Usually killing styles helps, but not always. My LONG version of the above css:
svg[style*="display: none"], svg[style*="visibility: hidden"], template, datalist{
display: block !important;
visibility: visible !important;
position: relative !important;
width: auto !important;
height: auto !important;
pointer-events: auto !important;
border: 4px dashed gray !important;
}
Ideally defined for certain domains only, if wrapped into @-moz-document.... rules.
Personally have it as global rule, but this only makes sense if javascript is usually OFF.
And be aware of possible false positives, of course this can just as well mess the look of some pages.
Can be added in userContent.css or adblock.css
.
1 -
1 line is enough, that's also usual :)
/* border: 4px dashed red !important; */
Just did some more experimenting, and delighted to find this hack really WORKS to easily disable/break/ignore any amount of stuff in my complicated 200kb css-file, no matter the content! Simply wrapping a whole bunch of code+comment blocks into new Begin/End markers like this:
@@comment{
....CSS to hide....
}
Even half the file or more without any worries about the engine "guessing" wrong End points 0 -
CSS COMMENT syntax
Had to look this up now, but the only valid comment syntax in CSS is still /*....*/
Although amazingly a javascript 1 line comment // seemed to work too when I now tested it. That would have been much handier. But after googling around, in CSS this is not treated as comment, but actually as error, and depending on the text content it may break next lines too. Pity!
This page explains it in detail: https://www.xanthir.com/b4U10
To sum up:
// 1-line comments like in JS:
Only SEEMS to work at first, but in fact are treated as errors, and linebreaks are not the end of it.
// => ignores single rules until next semicolon, or if placed before a whole ruleset, ignores it until next "}"
So // can work as lazy hack for RULES (2 chars // instead of 4 /* */), if aware.
But messy with free TEXT comments, still stops only at next ";" or "}"
A text comment could still be ended by typing a semicolon or {}, but that's already 3-4 chars again, no gain
HA! But some commenters on that blog pointed out that this could still be a brilliant hack to get NESTED comments! I often ran into that stupid prob:
a longer css code block with shattered text comments here and there, and then I want to temporarily disable the whole block, BUT: with normal syntax not possible. First had to edit every single comment inside too, a major PITA.
Someone also mentioned that it doesn't matter at all if "//" is used or any other invalid stuff, one uses "@comment"
=> Hehe, will now definitely start using this great trick!
Just not quite sure yet which error-chars to use exactly, needs a bit more playing. Doing a page search for double-slashes will find hundreds of hits inside URLs, but "@comment" could simply be put at beginning of a whole block. Hmm, but there's a catch, this breaks syntax-highlighting in NPP, still displayed like active code. Too bad... Or may simply use @@ instead of // to comment out just short single rules like @@border ? Will have to test this awhile... At any rate it's important to REMEMBER such hacks if used.0 -
TechnoRelic said:
> Ideally though, I would like the BNAVIGATOR Browser to 'pop open' that dialog
> where it ASKS you if you want to SET it as the DEFAULT Browser ...
Of course, but that goes without saying. It's not an open question whether or not that will "ideally" happen.
Don't worry, work on it is currently in progress (among other things), and just a matter of time until finished. Did you notice this was explained already? Actually, twice... or was it three times...
It's a bit strange that you keep pushing roytam, although you say you're happy enough using another browser in the meantime, so why... You want him to do double work? To hurry this up, as an urgent top priority? Perhaps he'll speed it up, or not, but for us end users it makes no difference whether it's done upstream or downstream ;-)1 -
MT: Well possible some of us are not the brightest here when it comes to computer science, but the fault for this misunderstanding here is your's. Not indicating what you're talking about exactly, using dev-speak, and clearly talking to RT in first post. Especially not bothering to mention to normal users who are not clairvoyant that "ShellService" might have anything to do with "Set Default Browser". That's absolutely not obvious, just a chance for blind guessing. Then again, it probably was your intention, to get more opportunities for jerking around.
RainyShadow said:
> As a last resort, you can try to use Control Panel > Folder Options > File Types
> and manually change the assotiations.
When using that method, this great little tool is an easier alternative:
https://www.nirsoft.net/utils/file_types_manager.html
One possibility for example is to open it and search inside where the current default "iceape" is used, then copy those lines for another browser. Am no expert for that stuff either, but guess at least four entries should be set:
For ".html" and ".htm" and "http" and "https"3 -
Great it's added in first post, thanx :-)
Just another example page:
https://www.chip.de/artikel/Pfannen-Test-2020-Bratpfannen-Testsieger-bei-OeKO-TEST_146932113.html
Now noticed that trick is especially helpful for a few pages, like cnn:
this method keeps the graphics and tables readable, unlike complete stylekill.
You're right that 8px red is too massive for general use.
I made it so extreme and with visible ##infotext## since my user css is so chock full with rules (incl. tons of notes 200kB) that they can interfere with each other. For example have cyan borders around all iframes, and lime green borders around all video stuff, and all sorts of red borders for various other stuff too, confusing myself what's what ;-)
Now fiddled and tested some more, and decided to make it just 4px too, like you.
Just not red, instead "gray". That's more intuitive to remember among my other styles (gray=hidden).
First tested 6px dotted gray, which looks best, but on sites where the text is glued to the border the dots are too hard on the eyes. Could add an additional "padding: 12px;" but better not mess too much, so returning to "dashed" again.
border: 4px dashed gray
Side note for tweakers: just learned today those css selectors must match the "computed" version too. For example an original style="display:none" becomes style="display: none;" and "display:none !important" becomes "display: none ! important;"
.
0 -
Can't believe no one (except 1) is ever using this very handy global permissions sheet, incredible.
But great it can be fixed!
While at it, in KMG74 it looks suspiciously the same:
with "Forget about this site" in the upper right corner, and the displayed values have nothing to do with my real global default prefs in about:config (most handled by macro, or just permanently blocked)
And if not too complicated (?), is there a way users of older KM versions might fix it themselves somehow?0 -
Just sharing a little CSS tweak for Retrozilla, K-Meleon1.5.4 and probably other browsers too, if you prefer to keep javascript disabled:
Without javascript, certain websites are either completely empty or crucial page elements are hidden. That can be done in different ways of course. One method is an SVG-trick for hiding content.
Examples:
https://edition.cnn.com/interactive/2019/05/europe/finland-fake-news-intl/
https://sourceforge.net/projects/jsminnpp/files/Uni/
https://notepad-plus-plus.org/downloads/
I suppose such pages will show up when javascript is enabled...
Without JS, they only show up when disabling inline-styles.
Retrozilla: View > Use Style > None
Retrozilla-Firefox2: View > Page Style > No Style
K-Meleon: Prob, that requires an updated stylekiller macro. Have a working beta,
but not ready yet for posting (another mini-macro grown to monster length, as so often, sigh)
But here's a better workaround, fully automatic and also without JS:
Just add a little css-tweak to your user styles.
Without domain conditions there may possibly be false positives too, but had no probs so far.
This CSS can be added in either
.../profile/chrome/userContent.css or adblock.css
(rename userContent-example.css if not yet done /
In KM adblock preferred since it can be toggled on/off during session, userContent not)
----------
/* NOSCRIPT workaround for some EMPTY pages, or incomplete content, if hidden by SVG-trick:
HTML is inside an invisible SVG-tag, requiring javascript to make visible.
Gray dashed border and ##HIDDEN## lines are OPTIONAL gimmicks, just to see where it applies.
*/
svg[style*="display: none"] , svg[style*="visibility: hidden"] {
display: block !important;
visibility: visible !important;
position: relative !important;
width: auto !important;
height: auto !important;
pointer-events: auto !important;
border: 4px dashed gray !important;
}
svg[style*="display: none"]:before {content: " (####HIDDEN SVG####) _ _ _ _ " ;}
svg[style*="visibility: hidden"]:before {content: " (####HIDDEN SVG####) _ _ _ _ " ;}
----------
Minor catch:
if you do have javascript enabled, this css is unnecessary, but can't add a condition "mark only if javascript is off" ;-) So IF you keep the optional dashed border and marker text, it will show up always, even if unnecessary.
If you want to apply this to SPECIFIC DOMAINS only, wrap this around the code:
----------
@-moz-document domain(example1.com),
domain(example2.com) ,
domain(example3.com) {
.....css code above....
}
1 -
Sigh, so much great stuff... which I'll probably not find anymore when needing it one day :-/
Just a little note regarding waybackmachine without JS:
archived versions can be found by testing various cropped dates in URL bar.
Without a starlet * in URL it needs no JS, and can not show a list, instead just opens the nearest hit.
For winimize.com this was the last available version:
http://web.archive.org/web/201109/http://www.winimize.com/
Later dates show the domain was for sale since 20120 -
RT-KM1.5.4+TLS1.2:
Wunderbar98 said: K-Meleon launches after copying both MSVCP71.DLL (v7.10.3077.0) and MSVCR71.DLL (v7.10.3052.4) from the formal K-Meleon v1.5.4 installer into @roytam1's extracted K-Meleon1.5.4en-US.tls12.7z directory, running k-meleonW9x.exe. It creates a new profile in C:\WINDOWS\Application Data. New unzip test with the two DLLs added plus a new empty file named 'profile.ini' launches the browser and creates the new profile in the extracted browser's directory.
Thank you for testing! Sounds great, to me like a 99,99% chance that a previous setup-version (impossible to download without unknown direct mirror link) is only necessary due to missing VB files.
So RT-KM154 could be made to work out-of-box if roytam1 would include those 2 dll's in the package.
And 2) make it portable with the ini-file, which all portable 7z KM-builds had since KM1.6.
And 3) if he'd compress it with an old zip-format that vanilla win98 users could uncompress without needing inaccessible SF-downloads, or better make it a self-extracting exe (which people like me who don't like exe's and already have a younger zip-tool could probably just unzip too)
I think it's really necessary that there's at least 1 of his browser builds (RZ-SM, RZ-FF2, RT-KM154) as "starter build" for vanilla98+IE5 users, needing no other resources of any kind. Which can then be used to further download otherwise blocked stuff from Sourceforge, github etc, needed for all further updates. Like KernelEx, or a younger 7-zip (v9.20 for Win98) which is necessary to unzip all his other browser builds.
End up using both RetroZilla and K-Meleon during sessions, can't decide yet which one i like better, both wonderful. If @roytam1's Firefox v3.6 worked in vanilla Windows 98 that would be magic. Loved that browser back in the day
FF3.6 (and current RT-build too) runs with basic KernelEx - you hinted somewhere considering it ;-)
It's really quite a generation leap from FF2 to FF3.6, and from KM154 to KM1.6.
For me most important, there sure are a lot more Firefox3.6 ADDONS as for FF2, let alone for Retrozilla-Seamonkey.
And in KM the slightly younger JS-engine matters too, for macros. While I hardly ever use javascript for websites, it's still needed for lots of site-tweaks and old-engine workarounds using little macro scripts, which very often use e.g. "querySelectorAll" which just didn't exist yet in FF2/KM154. Workarounds may be possible in many cases, but very bothersome. Just today struggled yet again several hours on amazon, grmpf, dabbling with a little macro script for article pages, to get those stupid links to other "colors" working again. Which are usually not about colors at all but for all sorts of product variations. Finally works now in KM1.6, but now wasting yet more time trying to add workarounds-of-workarounds for KM154? Ouf, grmpf, no thanks, took far too much time already...
By the way I finally realized I can test stuff again with "basic" KernelEx! Have a decade old version on my second disk, ancient backup, which actually still boots. It was even pre-release, KernelEx RC4, but don't dare to touch it anymore to upgrade. At any rate the crucial point about "basic" KernelEx is that it was much more stable, never gave me trouble with native Win98 apps, and was by default "off".
So did a few little tests now with RT-fx36.
Turned out it *does* start with basic old KernelEx compat set to "Win98"! But there's a catch, or rather two:
1) If explicitely set to "disabled", it crashes at start
2) xul.dll needs compat "Win2000", which KernelEx has automatically predefined.
Otherwise display is completely messed. Toolbars invisible and transparent, pages almost empty and no text either. Typical look for missing KernelEx.
0 -
I'm slightly shocked that invidious goes down the tubes now too :-(
And keepvid is unusable too again (or needs stronger JS / younger browsers)
Choices get slimmer and slimmer... all that's good gets destroyed since decades, what a shame.
(offtopic: and just discovered, the pageadviser domain is up for sale too)
Tested a bit yesterday, for me this seems the last remaining *working* download service - anyone know others?
(xxxxx = video ID)
https://bitdownloader.com/download?video=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dxxxxxxxxxx
Perhaps ruclip.com, but no chance without JS either.
Best chance IMO are userscripts which extract the full download-URLs from the video page source. Which is possible from all layouts, even polymer if the page shows only empty space and gray rectangles. Luckily that's what I'm using too.
The prob is, Google must allow this, or won't work. At the moment they still do, for those videos which officially have permission to be downloaded. But others (like new, protected music videos) are not downloadable/viewable this way at all. Am afraid it's only a matter of time until GoogleGod will decide he wants to block ALL downloads, which will be very easy. They could simply obfuscate all download URLs as they already do with protected videos, or worse, if visitors have JS blocked or don't fulfill other tracking requirements, not send them any URLs at all.
Without a download service and neither a userscript, the last resort could be to try and manually extract the page source:
Hit CTRL+U and search in html source for:
"itag\":18,
to find the download URL for mp4-video in 360p quality.
The URL is about 3000 letters long (!!) and ends shortly before "mimeType"
Other sizes:
22 = hd720p
18 = medium 360p
36 = small 144 (actually 320x240, save as *.3gp, only available with a MOBILE useragent)
.1 -
Wunderbar98 said:
Someone coming to this post fresh with a vanilla system should attempt the unzip only method first, then just update the thread, probably will work, it will be better confirmation.
Sure that might be "best", there's just a tiny little catch: That wait may take forever...
It's the year 2020 now. How many regular posters here still have a real Win98 system, not just a VM?
And how many of those, despite being on this tweaker forum, still have a 100% vanilla system?
And still of the rest (if there's any), how many never tried KM yet, which is often recommended for older systems?
Or perhaps a stranger, a passing reader with matching requirements may discover this topic in a few months or years, but doesn't have an account and will not reply. Provided he even cares enough to test diligently and report. And also provided he CAN get an account at all (I'd not be able to join anymore either, long since expired email accounts, thanks to pesky ciphers again)
That is, if such vanilla users even have READ access to this httpS forum, and not only IE5-6 and never find this. They'd first need to have an old Firefox on their system.
Therefore am afraid your own system is still the best vanilla chance by far in the foreseeable future.
In my opinion better a 99% reliable result now as a 100% sure result but perhaps never.
(Frankly, I also just don't imagine KM as a prog that does some mysterious behind-the-scenes stuff ;-) If I read you right, it didn't even copy the msvc-dlls into the Win system folder. Am not a dev but always had the impression it works very basic and oldfashioned, probably just looks up the needed dll's in either own folder or system folder, if global plugins are available, then checks if profile.ini exists to decide if profile folders are inside or out, and that's about it)
The K-Meleon loader was not installed or active, don't like these features. Figured out the issue creating program backups in the same Program Files directory, not a K-Meleon bug but a Windows Quick Launch "feature",
YIKES - never noticed! But you're right, just did a quick test by renaming a random program folder (have a few irfanview version folders), and bingo: the QuickStart shortcut updates the path itself when clicking it! Without any notice. Without rebooting even.
But you said your KM even starts itself automatically after reboot...?
Profiling firefox.exe with Dependency Walker, nice program, flags GDI32.DLL and KERNEL32.DLL. This probably makes sense, stock Firefox v3.6 never ran in vanilla Windows 98, stopped at Firefox v2. Failed launch of firefox.exe from a command.com window or click gets an illegal operation popup, stack fault in module KERNEL32.DLL.
Thanks for testing. And major pity. So RT-Fox3.6 still requires KernelEx for Win98 :-(
It doesn't mean much though that stock-FF3.6 never ran on vanilla98 either, since RT has tweaked this build not only to add TLS1.2, but also to work on older systems too. As he also does with a bunch of semi-modern browser builds backported for XP.
0 -
Thanks for investigating!
> Someone with a virgin K-Meleon vanilla system could test
Have some doubts if a "virgin KM" system is necessary, especially for testing portable versions. Perhaps RT-build would start working on your vanilla98 system after just adding those 3 files? (2 dll + ini) And then you could try again without msvcp71 for comparison, if it breaks again. There's also a slim chance that during initial tests you accidentally started it once with the normal exe instead of k-meleonW9x.exe, since the setup had installed the wrong target exe for Win98.
> Browser still launches on reboot
That's definitely not usual.... unless you enabled that old LOADER relic in installation options?
In the far past that thing confused quite some users in the forum, before it became useless anyway and was removed completely about 5years ago.
http://kmeleonbrowser.org/wiki/TutorialLoader
http://kmeleonbrowser.org/forum/read.php?1,132094
Have never used that loader thing, but if enabled, there should be an icon in the bottom right corner, with options on right-click. And a shortcut to loader.exe in startup folder, acc. Wiki.
Roytam1's Firefox 3.6+TLS1.2 (fx36) for ancient systems is here, but guess you tried it already:
http://o.rths.ml/gpc/files1.rt/fx36vc71-20171108_newnss.7z
It contains a "run.bat" to start it in portable mode, then it will use the already included profile, otherwise will start non-portable with the profile in global appdata or such.
The included profile contains a "xul.mfl", that's the startupCache of browser components, to improve startup speed. After major changes in extensions it's usually a good idea to delete it occasionally, then it gets refreshed automatically. But in case of startup probs not sure what's better, keep or delete it, would try both...
And interesting.... just notice RT-Fx3.6 comes with the same two msvc-files as for KM154 already included, but dating from 2006, younger versions as those in KM154-setup package (dll from 2003). Might be better....
And also just notice that my KM1.6, which matches the engine in Firefox3.5, contains the next generation msvc+r80 dll's, which require KernelEx. That supports again my previous impression that Fx3.6 is indeed meant for the same old systems as KM1.5 and Firefox2...
By the way do you know Dependency Walker? A very popular, tiny old tool to figure out which referenced system files are missing or buggy when a new prog doesn't work right. The screenshots may look a bit scary, but basically for us non-dev users all that matters are which files are marked red, as errors. Often just missing or outdated dll files. Mine is v2.2.6000 from 2006, and just see it's still valid for old systems:
https://en.wikipedia.org/wiki/Dependency_Walker
0 -
> Heck fun, most new (to me) software is exciting.
Hehe, sure understand. Just yesterday my whole evening was over in a blink after seeing this post from Jaclaz and testing a bit:
https://msfn.org/board/topic/181730-anyone-else-remember-vlb/?do=findComment&comment=1185185
A teeny, free, portable proggie Spread32, basically Excel light for Win98 - still can't believe it ;-)
Regarding choice, I used to think that roytam1 special Firefox 3.6+TLS1.2 build (fx36) would run on vanilla Win98 too.
Until it didn't for you. But then, RT-KM154 didn't work at first either, surprisingly, only after you first installed the old setup-version. Still wonder why, which additional files may be culprit. Would it have worked after creating a profile.ini switcher file, and copy over msvcr71 +msvcp71 from the exe-package, or possibly more needed? Any registry stuff perhaps, or anything? Especially interesting that you got an Advapi error, what I always regarded a hopeless killer message even with KernelEx. But this setup certainly didn't replace the advapi-file nor messed any kernelex stuff, so that dead end may have a magic workaround?!0 -
jaclaz said:
Should you want a small, simple, no bells and whistles, working worksheet running on *any* windows AND Excel/.xls/.xlsx compatible, check this SPREAD32:
http://www.byedesign.co.uk/
(portable, can run even from within a .zip)
That's insane. Completely incredible - a quite powerful EXCEL on Win98, as a teeny 3.5MB proggie (half the size being only language packs), free unlimited shareware with just a little nag, and portable! *dropdead*
It actually works. Of course, a few actions a bit awkward, e.g. editing multiline cells only glued together in the top input line, and sorting columns not quite as intuitive 1-click (finally found it in the menu), but for my very rare use that's still incredibly powerful. Thanks for the tip :-)0 -
> Nobody's ever managed to get a 9x command window to scroll
Except people with small screens, large fonts (10x18), and DOS window set to 50 lines - but only as "demo". Only allows to scroll through the last 50 lines, if the screen is smaller, that doesn't really count.
So the key to available scroll height is the window height, defined by the fix number of lines in the properties. Which is far too small.
As lousy workaround an output to a text file could be considered (xxx.exe > output.txt)
The menu option for scrolling only allows keyboard-scroll with arrow keys, provided the window does already have a scrollbar.
What I wonder:
acc web it's possible to set MORE lines in that DOS window properties, but probably only in newer systems?
> mode con: lines=9999
When I try that it gives only errors, except if setting the already available 25/43/50 lines as in properties :-(
Any hack to allow more lines in Win98 too...?
0
It was starting to become a real killer prob 1-3 years ago, half the web broken, even with basic KernelEx, but thankfully roytam1 appeared, and has posted a bunch of updated OLD browser forks! Adding modern TLS1.2 CIPHERS, and he also backported a few browsers to work on older systems again. That stuff is just shattered around here and there, a bit everywhere...
Title should somehow indicate this is actually about a DOS-browser with TLS1.2
KM-Goanna76 shares the same engine with NM27, and this "BigChange" build (which should IMO get a little version-number update) seems to have some severe new bugs:
Who's still using Win9x on the web besides me???
in Windows 9x/ME
Posted
Oh, and almost forgot, yet another TRAP:
Just because an updated browser may then be ABLE to handle modern TLS1.2 ciphers (more or less), doesn't imply that those were automatically enabled in user settings too! And not even, that checkboxes for those new choices would now appear in the GUI settings.
Depending on the browser, you may have to change the "security." prefs on about:config page directly. Or faster, filter for "tls"
The most important setting:
security.tls.version.max = 3 or 4
0=SSL3 / 1=TLS 1.0 / 2=TLS 1.1 / 3=TLS 1.2 / 4=TLS 1.3
Just as reminder, would create such an info pref manually:
security.tls.version.__INFO = INFO: 0=SSL3, 1=TLS 1.0, 2=1.1, 3=1.2, 4=1.3 beta
Depending on browsers, you may even have to add a list of new ciphers manually too - just which ones exactly...? Have only seen a few listed here and there in the forum, somewhere...
If anyone knows details, please post such a list here. Preferably in same format as other files for default settings, in .../defaults/pref(erences)/*.js
And except for too extremely old engines, Mozilla browsers have a page "about:support" which lists such stuff... "NSS" has to do with security TLS-versions... but that's all I know.
All that said, am no expert! Can't guarantee that all this is 100% accurate.
Just collecting tips here. If anyone has corrections, feel free to post!