Jump to content

Running Windows 98 in 2020 and beyond...


Wunderbar98

Recommended Posts

The useragentswitcher extension mentioned above works well, only issue is it defaults back to SeaMonkey default at browser startup:
Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.8.1.24) Gecko/20100228

The add-on has three extra built-in Vista-era useragents, the only one that still loads YouTube is:
Mozilla/4.8 [en] (Windows NT 6.0; U)

The add-on allows adding more useragents without re-working the extension. There are lots of useragentstring websites around, link below is pretty good, somewhat outdated, no JavaScript needed to view:
https://udger.com/resources/ua-list

The useragents i've quick added are:
Mozilla/4.0 (compatible; Dillo 3.0)
Mozilla/5.0 (Windows NT 10.0; WOW64; rv:40.0) Gecko/20100101 Firefox/40.0

A simple test is to load:
https://www.youtube.com

If the useragent is not supported YouTube will load a deprecated browser page.

Edited by Wunderbar98
Link to comment
Share on other sites


The resurrect_pages-1.0.7-mod.xpi extension from the sites linked above is only okay.

When right-clicking and selecting 'Resurrect this page' (or this link):

CoralCDN is defunct, don't bother.

Google Cache works well, using it frequently triggers a Google alarm.

Google Cache (text only) is good if you only need text.

Yahoo Cache function is broken. Yahoo still caches, was going to try and fix but Google cache works.

The Internet Archive function is broken, to fix:
- Unzip resurrect_pages-1.0.7-mod.xpi (its a zip file, 7-Zip works)
- Navigate to resurrect_pages-1.0.7-mod\chrome
- Unzip resurrect jar (also a zip file)
- Open resurrect\resurrect.js (WordPad may format better)
- About 2/3 down make the following change:

-->Before
case 'archive':
gotoUrl='http://web.archive.org/web/*/'+rawUrl
break;

--> After
case 'archive':
gotoUrl='http://web.archive.org/web/'+rawUrl
break;

- Re-zip everything, rename zip files to jar and xpi respectively
- Modified extension must have exact same directory structure, not nested directories
- Be sure to delete working directories before re-zipping
- Remove the old extension, re-install the modified extension

The Internet Archive is hit or miss, they do not cache as much as Google, sometimes the browser redirects to an Internet Archive donations page.

MSN Cache function is broken. Even on a modern Linux system with JavaScript partially enabled i couldn't load MSN's cached links from the Bing search engine, not worth the effort.

If knowledge is power the leader is obvious. The Internet Archive function may not be worth the hassle of fixing. The extension is only of limited use for accessing Google cache.

Link to comment
Share on other sites

For vanilla Windows 98 the biggest limitation of the resurrect_pages-1.0.7-mod.xpi extension is that it does not fix the inability to successfully load pages that use newer security protocols.

When attempting to load an inaccessible site, right-click selecting 'Resurrect this page' from the main browser window is useless as the extension does not recognize the URL in the address bar, just the failed about:blank browser window.

There would be merit in modifying the extension to read the address bar URL directly, or from clipboard, then load the otherwise inaccessible page from Google or Internet Archive cache. It could be another useful tool, in addition to the web-based proxy sites. I speak limited JavaScript but may spend some time on this later or just manually prepend the strings below as needed.

The magic URL strings, substitute $url with the actual full URL (eg. https://sourceforge.net).

Google cache
http://webcache.googleusercontent.com/search?q=cache:$url

Google cache text mode
http://www.google.com/search?strip=1&q=cache:$url

The Internet Archive
http://web.archive.org/web/$url

Note if The Internet Archive redirects to their donation page, even when running without JavaScript there is sometimes an Impatient? link towards the bottom of the page. Clicking the link will hopefully forward the browser to the desired cached page.

Link to comment
Share on other sites

addon resurrect_pages:
Interesting discovery! From your description this sounds very much that it could be the origin of an ancient K-Meleon macro, which I've just recently tried to update a bit too too!
To get past the TLS-blocks of course the retrozilla-veriations should help, and with KernelEx some much more modern roytam1-forks too or Opera12.12 (but didn't try as FF-addon yet, only as extremely simple KM macro)

Just wonder from your description, does the URL list not contain googleweblight yet?
First saw this extremely helpful resurrection proxy only in the same, extremely ancient KM macro myself:
https://googleweblight.com/i?u=$url
Just a quick copy of my macro notes:
Google weblight is a PROXY, pages are modified by Google to reduce content for 2G phone connections
Except: video pages, or cookie pages, or complicated pages....
INFO: https://support.google.com/webmasters/answer/6211428?hl=en
Older link was: http://googleweblight.com/?lite_url=http...
Attention: needs a mobile UserAgent! (or never stops loading?)
Attention-2: Google-Search now ALWAYS shows hits as weblight links if UA=Mobile, no escape!!
examples:
Mozilla/5.0 (compatible; MSIE 9.0; Windows Phone OS 7.5; Trident/5.0; IEMobile/9.0)
Mozilla/5.0 (Android; Mobile; rv:38.0) Gecko/38.0 Firefox/38.0
Opera/9.80 (Android; Opera Mini/28.0.2254/66.318; U; en) Presto/2.12.423 Version/12.16

And figured out the same fix for waybackmachine just recently too (/web/http), but sure could have direly used this trick several years earlier already!
Another wayback URL trick: if several date versions are stored, an older one can be accessed without JS too, by just typing any other old date, e.g. /web/200305/http... or even /web/1000/http... The server will automatically find and deliver the "nearest" archived date.
Great URL syntax info here:
https://en.wikipedia.org/wiki/Help:Using_the_Wayback_Machine
One HUGE prob I can't escape yet:
if the server has no stored version at all, it will immediately redirect to a new "save" link, creating a new version of e.g. a useless error404 page or addon-not-found page :-(
It's even possible to get a list of ALL stored wayback-pages with a direct API link, guess was something with "ctx" or similar in the name, but that's a whole different topic again.

Also created for my userContent.css or in KM rather adblock.css:

@-moz-document url-prefix(https://web.archive.org/save/) {
div.App-content-wrapper {background-color: yellow; border: 6px dotted red;}
}
@-moz-document url-prefix(https://web.archive.org/) {
a[href^="/save/"], form [action^="/save/"] {background-color: yellow; border: 2px dashed red; outline: 2px dashed red; padding: 1px;}
}
@-moz-document domain(archive.org) {
div#upsell-popup {position: relative !important;}
}


Link to comment
Share on other sites

Hi siria.

The resurrect extension could turn into something more useful, will play around a bit. I installed some Linux tools, a Bash script would be easier for me but may be clunky, as it would be seperate from the browser. This would, however, make it agnostic as everyone is using different browsers.

In vanilla the internet is fairly crippled. Without cached pages or a web-proxy at least 25% of pages fail to load. Getting >90% of pages to load would make the system much more functional online.

Regarding Internet Archive, the /*/ removal is good. The nearest date, interesting. Personally not interested in various archive dates, although others might like it. The /save/ redirect i've seen too.

First i heard of googleweblight, awesome and lightning fast, will incorporate something, thanks for sharing. Worked for me without issue on both Dillo and SeaMonkey v1.1.19 using default useragents:
Mozilla/4.0 (compatible, Dillo 3.0)
Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.8.1.24) Gecko/20100228

Funny, first read it to be google_we_blight not google_web_light :)

My Yahoo search engine in Dillo is good for showing cached links:
search_url="Yahoo https://search.yahoo.com/search?p=%s"

Interestingly Yahoo uses Microsoft 'Bing' caching. The cached link for a Yahoo 'ferrari' search:
http://cc.bingj.com/cache.aspx?q=ferrari&d=...

Whois bingj.com:
https://www.whois.com/whois/bingj.com

Forget to list the other Google cache URL earlier:
http://www.google.com/search?q=cache:$url

Thank you for sharing your notes.

Link to comment
Share on other sites

> Funny, first read it to be google_we_blight not google_web_light :)

LOL! That's brilliant xD Almost like they're hiding the truth in plain sight

Thanks for yet more interesting cache links, will check and add them to my macro too.

> The nearest date, interesting. Personally not interested in various archive dates

I think you will soon be, when discovering those are especially important for old systems. For example when the current page only contains information about their newest software versions, and all older stuff is long since deleted. Or the current page is only readable with javascript, or only with modern javascript, or as mentioned above, only shows "not found" or 404 errors instead of e.g. Mozilla's former "legacy" version of a Firefox addon page etc. etc.

> In vanilla the internet is fairly crippled. Without cached pages or a web-proxy at least 25% of pages fail to load.

Sigh, in my impression meanwhile rather 60-80% of pages broken, not because of the system, only due to missing modern TLS in older browsers! :-((
Especially when it comes to researching any current stuff, like news or products. E.g. yesterday was googling around for smartphone models, and really most pages didn't even show up. Then have to fire up one of roytam's fallback browsers, which are lifesavers, but cannot use those as main browsers, either being not my beloved KM or too buggy yet.

Speaking of googling:
Just the last week google killed the classic search view too, yet another huge desaster :-((
At the moment only the normal page, while the image search still works, but sure will be next soon.
No one else a prob with that?
Had to waste two evenings again just struggling to figure out little css+js snippets, to get the results view bearable again (in KM1.6 without JS). And now have to toggle the useragent back+forth all the time too, for googling, grmpf. And found no way yet to figure out the number of hits without JS, argh.

Edited by siria
Link to comment
Share on other sites

Hi siria.

Hiding in plain sight, maybe yes. At first i thought it was actual competition, disappointingly not. I don't speak Chinese or Russian, wonder how their search engines fair.

Depending on my mood, sometimes wasted hours, sometimes a challenge. If your head gets sore from banging against a wall, eventually it's time to stop.

My 25% breakage estimate was conservative, yours is probably accurate. It's frustrating to navigate the web and often too much trouble to repeatedly load a web-proxy. Plus bookmarks created while using the web-proxy don't work.

It would be great if a URL could pass through a web proxy without actually visiting the proxy page. Or if a web-proxy could be set up as a permanent browser proxy, but i have yet to find anything like this.

I've been testing proxies from the site below in SeaMonkey without success, hoping a less secure proxy might just pass through data as HTTP only (no HTTPS). JavaScript is needed to get the IP Address. Even selecting Anonymity: None fails on pages like https://sourceforge.net. Many proxies are unresponsive.
http://www.freeproxylists.net/

Most search engine results obfuscate the target URL, making it a hassle to cut/paste the raw URL into a web-proxy. Google search result link example for Ferrari Wikipedia in SeaMonkey:

http://www.google.com/url?q=https://en.wikipedia.org/wiki/Ferrari&sa=U&ved=2ahUKEwiglLaLtZXmAhVUoZ4KHaKJBgYQFjAAegQICBAB&usg=AOvVaw0fHm2OAw5XYLRBnrqJ9Eo5

If a site does not load, mostly i rely on cached links or just go to another site, which is limiting.

You're probably right about archived dates, i will research more, thanks for the earlier link.

K-Meleon may be the best browser standing for this old OS, i'm used to Dillo and SeaMonkey.

I don't use Google search much but did notice a change. Often in Dillo, the Google search results no longer provide clickable links, now useless. For me Yahoo with cached links is the best, even without JavaScript. Most search engines require JavaScript to access the cached link.

My favourite startpage search will probably never work in Windows 98.

Link to comment
Share on other sites

It's been a longer while that I tried other search engines, but always found they are lacking important info, so was forced to stay with Google. Although strongly disliking the spying, but can't help it, and yahoo&Co are probably not much better either. While on the more private engines like DDG, IIRC was especially missing the page date. Although, meanwhile most google hits have no dates anymore either. And the other crucial goodie was always the direct cache links - which are now also gone of course. Aargh. But at least can open them now by menu, was not aware that was even possible until recently. The only catch is that not all pages are cached, what can only be figured out now after trying to open them.

The obfuscated search links don't catch me, hehe - when using a middle-click to open them in background, what I've always done anyway, one of my macros reads the LinkURL and cleans it (if necessary) before opening the new tab. Am sure there exist also FF addons for that same purpose, have seen them in passing, but can't remember details.
What's surprising me now is that you say you often get no clickable search hits at all anymore, very weird.

But overall I really don't want to use proxies all the time. For occasional use G-cache and G-weblight are okay, but when it gets too much, still prefer to fire up a fallback browser.

Link to comment
Share on other sites

In regard to URL redirections there's an old Firefox (v10-v23)/SeaMonkey (v2.7-v2.20) add-on called Redirect Cleaner that I've been using for years in Firefox and Pale Moon (slightly modified for the latter to include its ID for v24-v28). Still using it (v2.2.0) in Pale Moon 28.7.2 under Linux Mint.

For a long time I've been using QuickJava, an add-on that can enable/disable JS, Java, Flash, SilverLight, Cookies, Images, Animated images or CSS with a single click. It can sit in the StatusBar or in the Toolbar. My current version 2.0.4.1 is available only for Firefox (v20.0-v29) but it works in Pale Moon too, and I did have an older version installed in Firefox 9.0.1 under 98SE.

Something very useful to me has been YARIP. It's very powerful so it's easy to make a mistake. It can permanently remove unwanted elements from a web page and it can also alter its aspect through CSS. Current version 0.3.4.2 is available for Firefox v17-v28, Pale Moon v25-v28 (added by me), Thunderbird v17-v28, SeaMonkey v2.20a1-v2.25 and mobile Fennec v17-v28.

For such old add-ons there is another add-on called Classic Add-ons Archive (v2.0.1), but this one requires high Firefox versions (v45-v56) or Pale Moon (v27.0-v28), while minimum SeaMonkey version is 2.40. Maybe someone with a dual-boot at hand could install it in a newer OS/browser and download old add-on versions compatible with 9x browsers, then make them available somewhere.

I built myself an AHK script that removes certain redirections from URLs, it works by monitoring the Clipboard, so it can work with any browser, e-mail client and pretty much any online/offline documents as long as one can use Copy/Paste. Anybody could build such tool if needed but it does require some programming knowledge.

Link to comment
Share on other sites

Hi siria. Did quick Google tests in Dillo. The issue is only partially useragent related, there appears to be more breakage going on, maybe just a Dillo thing. Also Dillo does not use JavaScript, suspect modern search engines do not like this behaviour.

In Dillo a modern useragent now provides cached links but many regular links still remain non-clickable, even featured advertising. The key may be to find a Goldilocks useragent.

On old SeaMonkey's default userstring without JavaScript, Google provides clickable results as normal but no cached links. So much for an open, accessible to all internet.

Hi Drugwash, thank-you for the information. AFAIK the browser versions mentioned in your post left vanilla Windows 98 behind a long time ago. The add-ons mentioned may be good for those who can modify them for older browsers.

Maybe not much new under the sun, just created a prototype helper script too for internet related issues. Presently it also reads/writes to clipboard and i hope to add a URL strip function later.

Just curious Drugwash, did your clipboard monitoring script run as an automated daemon or how was it interfaced for user interaction?

Link to comment
Share on other sites

Cygwin is a Unix-like command line environment that has not supported Windows 9x for years, not even Windows XP. They provide archived setups that run on Windows 98 but no working legacy repositories could be found through the time machine. Cygwin-lite works, will add a tutorial soon. If anyone knows how to set up a full blown Cygwin install on vanilla Windows 98 please let me know.

Edited by Wunderbar98
Link to comment
Share on other sites

Recently upgraded 7-Zip from v3 to v9.20. Works well and can now extract tar.xz files. Nice 7-Zip has provided long term support for older operating systems. Not sure what version they are on now, or even if this is the latest version working for vanilla Windows 98. Good enough, it's often hard to find downloads when many websites are not accessible.

Still playing with SeaMonkey extensions. May as well test and grab anything desirable before these old sites go down. All SeaMonkey / old Mozilla add-ons discussed were found on the add-ons links provided quite a few posts earlier.

Dropped resurrect extension as it was not useful, commented on in earlier post, will be replaced by a custom script.

Added paste_and_go, now a standard feature of full-featured browsers to manually load URLs.

Added quickjava-0.4.2.1-mod.xpi. A much older version than mentioned by Drugwash a few posts up. Seems good, i don't use Java but it makes it easy to enable/disable JavaScript from the status bar. Have not yet had time to test what else this old version can block.

Added Resizeable_Textarea_0.1a-mod.xpi. Allows resizing text boxes when replying to forum posts, not wider just longer (on this site anyway), semi-useful.

Tested video_downloadhelper-3.4-mod.xpi, seems broken for YouTube, did not test other sites, too good to be true. Will find an alternative solution for vanilla Windows 98 that does not require JavaScript or Flash.

Edit:
Also added undoclosetab.xpi, also now a standard feature on modern browsers. Didn't realize how often i rely on this.

This old SeaMonkey is starting to feel a lot less clunky.

Edited by Wunderbar98
Link to comment
Share on other sites

Learned more about removing extensions from old SeaMonkey v1.1.19 but can't find the link. Manual extension removal is acceptable as there is no about:addons functionality. There was apparently a 'removing extensions' extension but i could not find it.

Installed extensions and *.rdf files are stored in the profile's chrome directory, default path:
C:\WINDOWS\Application Data\Mozilla\Profiles\default\your_profile_name\chrome\

Recommend backing up the SeaMonkey profile, or at least the chrome directory, before surgery.

As prevention is better than cure, restoring a good chrome directory after installing extensions that may not be desired is easier than having to remove them later.


Manual Messy Method

The manual extension removal method is messy and incorrectly modifying *.rdf files causes breakage. These files, however, accumulate cruft when testing and removing extensions. For example, a fresh chrome.rdf file is 2 KB which quickly bloated to 38 KB after testing extensions.

1. Manually delete the *.jar file related to the applicable extension.

2. Use WordPad to delete applicable lines and sections related to removed extensions in the following files. As this is where breakage may occur, it's okay to skip this step if the cruft in these files is acceptable.
- chrome.rdf
- overlays.rdf
- stylesheets.rdf

3. Restart browser, if breakable occurs restore profile and try again or see below.


Automated Clean Method

This is a cleaner and more automated. After testing extensions and deciding what you want, reset the chrome directory and re-install only keeper extensions. As the browser is outdated, new extensions and updates are not forthcoming anyway.

1. Delete the entire chrome directory, which removes all extensions and configurations.

2. Restart browser, which re-creates a fresh chrome directory.

3. Manually re-install only keeper extensions.

4. Restart browser, nice and clean, everything should work.

Link to comment
Share on other sites

On 12/3/2019 at 10:14 AM, Wunderbar98 said:

Just curious Drugwash, did your clipboard monitoring script run as an automated daemon or how was it interfaced for user interaction?

It's an independent aplication, a compiled AutoHotkey script that runs as any other application and captures the Clipboard operations. When anything is copied to the Clipboard it analyzes the contents and if it matches any of the predetermined patterns it processes it and copies the result back to the Clipboard so the user can simply paste the processed result to a browser's address bar, a text file or wherever they want.

On 12/3/2019 at 10:14 AM, Wunderbar98 said:

The add-ons mentioned may be good for those who can modify them for older browsers.

That's precisely why I mentioned them. Older versions of those add-ons may be found to work with 9x-compatible browsers or newer versions could be modified for 9x compatibility. Either way, some people may just want to find out about them regardless of their OS version, as they are really useful IMHO.

Link to comment
Share on other sites

Hi Drugwash, thanks for your reply. Your AutoHotKey script sounds great. I reviewed an AHK tutorial, looks ideal for Windows 98. Wish i learned something like AHK or Visual Basic before.

As it sometimes takes a lot of hacking to get an extension modified, shame most of these become custom one-offs. Just daydreaming about an organized central Windows 98 repository full of hacked goodies. Everything seems so fragmented now.

For anyone getting started with extension hacking, this is something that worked well for me with old extensions that install out-of-the-box. First install the original extension, then just unzip the extension's *.jar file in the user profile chrome directory, hack away, re-zip into a replacement *.jar file, restart browser and test the modifications.

Only re-zip everything into a proper *.xpi extension when it works to your liking. Much faster than un/zipping and formally removing/re-installing the entire *.xpi extension for each quick test. Maybe this helps someone save some time and hassle.

Reviewed Windows 98 minimum system requirements, though probably slow, crazy it runs on a 486 with 16 MB RAM:
https://en.wikipedia.org/wiki/Windows_98#System_requirements

Kudos to those who still tune and tweak this OS and/or use it as a daily driver.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...