siria

flyod said: > just a while ago, i realized i can also force the window contents to update by moving > another window in front of it, then clicking back to the palemoon window. > also, if it would help narrow things down, chrome-based browsers has a very similar problem. Glad KM runs again. And no idea, am no expert at all, but could it be the browser is moved by the system into background and some invisible task jumps into foreground? Perhaps some cleaner, indexer, or malware, system plugin, whatever. But just wild guessing. Or a driver prob as you wrote here: https://msfn.org/board/topic/180462-my-browser-builds-part-2/?do=findComment&comment=1182803 Does it happen too if you disable javascript completely? Or set the browser offline? If you save a little static page to disk and open this one only? If browser related, a couple more prefs which may or may not be involved, trying can't harm: layers.low-precision-buffer dom.ipc.plugins.asyncdrawing.enabled = false other ..."ipc" prefs... layout.interruptible-reflow.enabled (true or false?) plugin.scan.plid.all = false kmeleon.plugins.external.load = false browser.sessionstore.max_tabs_undo = 0 javascript.options.jit or ".ion" prefs or "wasm" = false ..."prefetch" prefs (Create manually: "network.dns.disablePrefetchFromHTTPS") network.predictor.enabled profiler.enabled (telemetry collection) ..."workers"... ..."webcomponents"...

flyod said: > i'll also redownload on the off chance maybe the download just didn't come thru correctly Ah right, thanks for reminder! I assumed you'd been trying KM since awhile, but if build 20200613 was your first attempt, that's the culprit. It was first crashing on everyone, until roytam1 fixed it just half a day later: https://o.rths.ml/kmeleon/ (as usual keeping the old date still in filename)

flyod said: >i feel like if i could figure out the reason for the immediate crash of k-meleon, > it might get me close to finding the problem with everything. is there any way > to enable some kind of logging mode apart from just the windows erorr reporting? When does it crash - at startup, or when loading pages? If at startup, did you customize anything or never even get that far? If not already done, try if helps to delete the StartupCache folder in profile. Or try a new profile, e.g. by toggling from portable to non-portable mode (renaming empty file KM/profile.ini to profiles_renamed.ini or delete it completely, or if it doesn't exist, create it) Occasionally it happens that KM tries to restore a session with some buggy page and crashes. Can be all sorts of things, or in the past was also caused e.g. by russian URLs (cyrillic?), and someone else got crashes due to a corrupt bookmarks file, or e.g. 2018 (Firefox) kept crashing on someone who then figured out he needs to set pref "browser.display.use_document_fonts"=0 (disabled) to overcome a buggy page in session. Or hardware related prefs.... or some prefs for plugin-container, for gfx or opengl At any rate KM/kplugins/update.dll can be removed. It had an old bug that could crash, hope meanwhile fixed, but for RT builds it's completely useless anyway. And IIRC KMG76 needs SSE2 (or was it only SSE?) Or a KM-user once mentioned he needed to install VisualBasic 2005 (or is that 2010 for current?) to stop startup crashes. KM had an ancient crashreporter plugin, but no idea if still working: http://kmeleon.sf.net/files/crashrpt.zip It needs that pref created and set true, also possible by manual edit of prefs.js: kmeleon.plugins.crashrpt.load

Thanks for the youtube links. Yep same for me, all 3 are forbidden to download. Meaning the yt-script contains only the various download properties but the URL itself is obfuscated. It's probably possible with some sophisticated tricks, but probably a bad idea and anyway over my head. Oh well.

Wunderbar98 said: > Came across a site yesterday that can download YouTube videos via webpage. > It does not appear to get them all, had a VEVO video still fail, but seems to get most, > many of which would be inaccessible to my 9xweb script. Thanks for the tip :) Awhile back tried to get a video as mp3, but no chance. Next time will try keepvid. Just curious if my own scripts works with those 'inaccessible' vids, if you can post links? But can only download allowed videos too, not protected ones. (PS: sorry but I cannot run cygwin or anything, zero chance)

Talking of bandwidth... Not sure if this helps anything, and perhaps mentioned before, just throwing in what very little I know. There exists also a small 320x180px (funnily called 144p) version as 3gp. About half file size as medium 360p and of course quite fuzzy. Can play it with old MPC. Available only on (all black) mobile "m.youtube.com", with e.g. the ipad-UA someone posted here recently Mozilla/5.0 (iPad; CPU OS 6_0 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A403 Safari/8536.25 or IE9-mobile UA: Mozilla/5.0 (compatible; MSIE 9.0; Windows Phone OS 7.5; Trident/5.0; IEMobile/9.0) or FF38-mobile: Mozilla/5.0 (Android; Mobile; rv:38.0) Gecko/38.0 Firefox/38.0 UA iPad is better since it contains 3 formats incl. 720p, while others gets only 360p+144p Hmm, could swear IE9-mobile was getting streaming RTSP-links last time I tried... Or was it 240.mp4? Where has that one vanished now...?! Anyway, trying to find rtsp again, by UAs, OperaMini still gets it, but ONLY in 144p/3gp: Opera/9.80 (J2ME/MIDP; Opera Mini/9.80 (J2ME/22.478; U; en) Presto/2.5.25 Version/10.54 HUH.... and although equally called 144p/3gp, this rtsp-version is dramatically smaller as the (fake) 144p/3gp above?! Probably not watchable at all, except for preview checks... Weird, and the screen ratio is totally different too, although from the same video. Don't get it again, oh well - as usual with video stuff. 144p in FF38 or IE9-mobile versus OperaMini: (http) small 144p = 320 x 180 px 2.32 MB 226 kbps VIDEO 3GPP (mp4v.20.3) (itag:36) AUDIO_LOW (mp4a.40.2) (rtsp) small 144p = 176 x 144 px 0.76 MB 74 kbps VIDEO 3GPP (mp4v.20.3) (itag:17) AUDIO_LOW (mp4a.40.2) (or is this too offtopic here? Can delete it again, no prob)

Bruninho said: > However SM still crashes A LOT with JS enabled. I have even disabled > hardware acceleration for the browser, no luck. So I disabled JS for now. > I was testing SeaMonkey 2.32, quite stable if you ask me now. Just don't run JS. > For me this is strange. I know the browser can understand the jQuery calls (most of them) > and modern JS. I still think it's a caching related problem, either that or the OS can't handle > the amount of scripts involved in any website. If that means you can't enable ANY javascript at all, it might be worth testing some partial blocks, not only all-or-nothing. Yes I know that will also partly kill functions, but better as none at all. On about:config page, would start with toggling off prefs containing: "jit" (for older engines until 26) ".ion" (younger engines since 27, note the dot to reduce hits) and all "workers" prefs Further tests perhaps try blocking: default content block permissions are hidden out of box, must be created manually on aboutconfig (as INT). Values are 0 or 1 (all allowed), 2 (all blocked), 3 (only same domain) permissions.default.script (only affects external scripts, loaded from a "src" link) permissions.default.xmlhttprequest permissions.default.xbl permissions.default.indexedDB or same prefix for ...object...ping...beacon...font... dom.storage.enabled svg.enabled indexedDB reminds me that SQLITE is completely broken in my KG74, absolute KILLER for history and lots of other crucial stuff. But my main KG74 version is probably not the latest one by RT, only tried newer ones shortly and found they don't help, so postponed due to time probs.

My old laptop is just standing on additional, extra high rubber feet. Occasionally it gets a bit too warm for my taste, then switch on a little USB table fan, standing netxt to it. Blowing from the outside in, just sideways. No idea where the air gets in, and not much obviously, but glad it seems to help too ;-)

schwups said: > The "ca-archive-2.0.1.xpi", min. version set to 24 in the rdf file, gives a unusable, blank add-on page on FF35. > A reboot to XP and running it on FF52.9 is very inconvenient. Yeah, same prob in KMG74 :-( It's really stupid. I do keep suspecting it could easily be fixed for older browsers too (how?) by probably just adjusting a few lines (which?), but no addon experts here to help. Sigh. A few addons are still left here, but for Firefox only those that were also compatible with Seamonkey: https://addons.thunderbird.net/en-US/seamonkey/ Only mini-descriptions and no version info, but quick and handy to load: https://web.archive.org/web/20191102092539/http://legacycollector.org/ https://web.archive.org/web/20191102092539/http://legacycollector.org/firefox-addons/index.html Firefox addons in waybackmachine: Prob: seems to be stored only partly, and SEARCH field doesn't work https://web.archive.org/web/20181002055204/https://addons.mozilla.org/en-US/firefox/extensions/

Jody Thornton said: > Just remember that Waterfox has been purchase by a search/advertising company, > so even there I'd be weary of its future. Yeah it's a horror :-( Especially seeing that it's the source of a FLOOD of imported commits in roytam's NM browsers :-( And aren't they also the (only) hoster of the invaluable 15GB xpi-files of Classic Addon's Archive? :-( The only archive of FF xul addons left, after Mozilla deleted them all and awhile later even legacycollector was killed by attackers. So now, with Waterfox going downhill too, there's soon only waybackmachine left for all those precious addons. But who knows for how long yet until being forced to remove them there too :-(

Tripredacus said: > I had talked to Xper last week about this issue. He says that there are some > IP ranges that are blocked due to Google Adsense abuse. It effects a lot of > countries including, at least, Poland. Whole MSFN domain is BLOCKED? Since MONTHS? In LOTS of countries?! Oh no :-( But I'm not buying that this were only "accidentally", otherwise it would long since be open again, not obviously permanently :-( Suggestion to update the title accordingly, sounds way too harmless for such a desaster.

@Bruninho website Just some thoughts, since you seem to be really interested in old systems/browsers. I understand we're coming from opposite ends of the spectrum, and your site is meant to showcase "modern" features of course. But you also claim it were "responsive and efficient", and sorry, but I have some doubts there. It may be true for users with modern systems and fast connections and no bandwidth probs, but that's not the target group which needs efficient pages. On less-modern equipment the site is instead rather heavy and unresponsive. The #1 killer, and I mention this because it may matter for some customers too: the site even denies access *completely* to all sub-pages if readers can't use javascript for whatever reason! (limited resources, hardware, bandwidth, system, security, whatever.) Of course some advanced functions may need JS, but that doesn't imply that the whole rest need be broken too. For example look at this forum here, or look at Amazon: their pages surely have tons of functions, and (I suppose) also stuff requiring modern browsers. Yet even for browsers as ancient as FF3.5 and slow connections it's still possible to load them in a reasonable time, and without needing javascript at all. It's possible. On amazon JS is only needed for secondary functions, like reading comments of comments, since a while. But browsing and searching still works okay without JS. Of course, partly bad layout in very old browsers due to CSS, but that's secondary, what matters most is that a page is still readable at least. On your homepage only the entry page loads normal -although slowly, and page width only 15cm on a 28cm wide screen- yet none of the subpages and none of the language variations are accessible at all. Completely blocked. No go. Or something I never understood when stuck on ultraslow connections in the past: the images look small, but in reality are already loaded full size. That costs a lot of unnecessary bandwidth (your page size 3MB needing ca 10min), and never understood the point: if only shown small, why load them big? Or otherwise, considering they are already loaded full size anyway, a nice gimmick could be to show them full size just when hovering over them (css img:hover, no JS either) That said, next thing I'll do now is open my user css and add this as a global rule, LOL! Sigh, just noticed killer #2 for old browsers: the whole browser crashes when loading your page with SVG enabled.

Have sadly lost track again, so much precious info here but all faaar too much to memorize and over my head anyway, just want to mention YOUTUBE: there's also a youtube downloader script that works in K-Meleon, also in Win98: http://kmeleonbrowser.org/forum/read.php?9,142050 The old version was the great old userscript by Gantt, for all sorts of browsers. Over time it got a few little update-fixes by John after youtube changed their code, and after their last major rewrite he decided to simplify the whole thing to keep only the most important functions, matching his own needs. And also to get the file size small enough to work easier in old KM-versions too. This current downloader script also still works nicely in my old KM1.6 (like FF3.5) The script itself is plain javascript and surely works in other browsers too, just depending on injection methods (addons?) the header stuff may need a check again. Additionally am experimenting with my very own script, started from scratch, which now downloads from all YT layout versions incl. mobile, showing all available quality links (my impression, not fully sure). Looks like only the initial element IDs are different but the rest of the downloader code can remain the same for all layouts. Downloading works fine, but currently on backburner as unfinished early beta, so no upload planned in the near future (not that anyone would need this anyway) What I had NOT realized until just recently though, until Grey_rat (IIRC) here posted some useragent examples, was that the MOBILE layout delivers different file versions too, depending on UA! My usual old UA had only delivered rtsp:// links, so was very happy to learn that now tiny 3gp and 144-mp4 versions are available again too, just depending on UA. Thanks :-) What's confusing me a bit now is to read about old K-Meleon75RC3. IIRC that's a rather old dev KM version by Dorian, pure gecko at the time. Did roytam post some update for it too?? Or if not, wondering why using a prelim RC3 version, and not the stable version released a bit later. Well possible this was explained here before, have lost a bit track as mentioned ;-)

cc333 said: > I've found this forum to work quite nicely on both Opera 12.02 and @roytam1's special > Firefox 3.6 build, particularly if I switch the forum theme from "Switch (Default)" to "IPS Default." > "Switch (Default)" is usable, but kinda glitchy with the odd layout issues that people here have reported, > whereas "IPS Default" renders almost perfectly (although the reply box is broken on Firefox) Goodmaneuver said: > I can confirm the IPS Theme fixes width issue in Opera and Mozilla cannot reply Confirm too (but checked only width) So that's why some people got this verry narrow view and some other people not!

No april I think ;-) Okay can't remember that anyone got FIREFOX running, but the engine inside worked since years, wrapped into KM-shell. So perhaps modern KernelEx versions are now able to run pure Firefox too? No idea, but who knows. As jumper mentioned above, he had gecko31 already running in 2014 too, with the same result: it's running, but very slowly: https://msfn.org/board/topic/172801-new-k-meleon-74-version-now-listed-at-various-download-sites/?do=findComment&comment=1088397 Although the KM-number 74 normally indicates gecko24, this special KM74_31 was an early experimental build with gecko31 inside. And of course we also have roytam's special KM-Goanna74 (with engine 24 inside) build for Win2000, with updated TLS1.2 inside. Running in Win98 with a modern KernelEx and compat set to Win2000 https://msfn.org/board/topic/177485-backporting-newer-browsers-to-win9x-with-kernelex/?do=findComment&comment=1175954 http://rtfreesoft.blogspot.com/2020/01/ THIS would basically be very much usable on 98 with modern Kex, if only it weren't so very buggy yet. Am sure that's only partly due to the system, but partly just general and fixable browser bugs, already contained in the original builds. Hmm.... could almost bet someone (jumper??) has posted awhile back that he even managed to get KMG76 (with engine PM27/generation mozilla38) running too on 98, with modern KernelEx! Just where was this, can't find it at the moment.... HERE! Just a few pages back here, on page 8, last October: jumper said: > I now have KG76 working again. Pages render much faster than KG74, but scrolling lags and menu text is mainly garbage. https://msfn.org/board/topic/178283-how-you-really-browse-the-web-on-98me-in-2019/?do=findComment&comment=1171276

youtube-search: just tried with my ancient browser, and page was empty except a view grayish rectangles. Then tried at random with another useragent string, and the page showed up again. Mozilla/5.0 (Windows NT 6.1; rv:38.0) Gecko/20100101 Firefox/38.0 Attention, some browsers come with a bunch of predefined site-specific useragents, that includes domains for youtube and googlevideos. Can be changed on "about:config" page. You can set such strings empty to make them get your current global UA string (true default string or a global override UA)

loblo said: > Not sure which site specifically you're looking for but Opera 12 does great with > https://www.worldometers.info/coronavirus/ through https://sitenable.asia/ web proxy. Thanks! YES this site shows curves in Opera12. Was only confused about the proxy, don't need it for that site. But will come in handy for other sites. What I'm mostly after are curves of total infections in some Western and Asian countries, separate curves but together in 1 diagram, for easy comparison of development. Have seen an old version somewhere as screenshot, and in the upper right corner was a button to "add more countries", but couldn't get it working at the time. (offtopic) Side note: if anyone is wondering about the "mysteriously" low death rate in Germany, that becomes slightly less mysterious when knowing that hardly anyone "qualifies" to get a TEST. The normal population isn't aware of this, and the conditions may vary depending on which area or federal state people live in, but after reading countless news around the web for weeks, have also come across several announcements that in order to save the far too few and precious testkits, only those people are allowed to get a test who: - are already SERIOUSLY sick - AND have visited superdangerous areas, or had close contact with an "officially" infected person - AND are extra endangered (high age, previous diseases, healthworkers) That 3-conditions requirement was stated by official health institutions. Otherwise no chance for a test, no matter how sick, and if any possible yet, non-qualifying people are simply told to stay at home and isolate themselves, just in case. Was also confirmed by several forum or article comments of people who tried hard, but no chance. And of course, people who die untested, e.g. in retirement homes, die of "normal" lung infections or heart probs and will never land in the statistics either. Made me wonder about how anyone ever got tested, but guess the final decision is still up to people's family doctors. The most incredible thing I've read, in an interview some official person actually declared something along the lines of "infection numbers are now just estimated, it's a normal procedure worldwide, and more accurate anyway as just a few tests here and there". But saw this only once, and just one sentence, so who knows. Would sure look like them though. But considering the test-requirements it just keeps amazing me again and again how they get off with shiny claims of "early mass testings", but oh well, politicians. Only a few days ago the the number of conditions was officially reduced from 3 to 2. Except that now they're running out of "chemicals" for testing, and can do less tests yet again.

Anyone know how to get those **** virus-diagrams visible in old browsers, or yet better, a site that shows them as normal images?? Everywhere I get to see only teeny tiny preview thumbs now, but no readable CURVES :-(

(slightly offtopic) Speaking of transparency, there's an amazing tiny image editor named Pixelformer, which can show bitmaps with partial transparency in Win98. For example the toolbar buttons of the Privacy Bar in K-Meleon, in default skin AuraAlpha, look out-of-box just black-on-black in Win98. That's due to the way partial transparency was stored in privcold+privhot.bmp. Even in other image editors it's near impossible to see readable text on transparent background, only Pixelformer can do it! At first look it blew me away, an image editor with a download size of 0.9MB, unzipped still just 2.6MB, superfast - but with layers and all sorts of other advanced features! At second and third look, sadly, found it's only (??) silver not gold, too many important features still missing. But nonetheless, highly recommand to grab it while still available. Have downloaded it 2009, now checked and luckily the homepage is still there. And still last version 0963RC3, stable free beta version. It's a setup file, but the exe can simply be unzipped manually, no installation needed. And the page claims it needs min XP, but seems to work just as well in Win98 (my KernelEx compat is "Windows 98 SE") http://www.qualibyte.com/pixelformer/

No the zip does not contain those files, but if you run update. bat (if I read it correctly?) it copies the 2 exe files from the zip to your c:/Windows/KernelEx folder renamed to dll. That's why I meant to check inside this folder. This renaming is probably just a typo. Except if that was really done intentionally, but doesn't seem so since the blog maintainer replied something like 'oops' ;-)

@Bruninho ...and if using advanced KernelEx from that blog you linked, still with the typo in update.bat (since 2015?) compare the files in the downloaded zip with those in your c:/windows/KernelEx - If you find "ApiLog.dll" and "Verify.dll", delete those - Check if properly copied "ApiLog.exe" and "Verify.exe", with latest dates

Bruninho said: > Thanks for the tips about the userContent.css! I will check it out with RetroZilla. And here you're already running into absolutely crucial css features MISSING in Firefox2. 1) "domain" probably OK? @-moz-document domain(bleepingcomputer.com) { header {position: relative !important;} body div {background-color: yellow !important;} } 2) but "url-prefix" needs min FF3: @-moz-document url-prefix(http://example.com/zzzz/yyy) { 3) and "regexp" needs min FF6 acc moz-dev: @-moz-document regexp("https:.*") { https://developer.mozilla.org/en-US/docs/Web/CSS/@document https://developer.mozilla.org/en-US/docs/Web/CSS/At-rule also stuff like css selector ":not()", needs min FF3.5

Bruninho said: > IMHO a very good web designer should take it in account when > designing a new site (backwards compatibility) Oh yes, fully agree of course :-)) > I wonder if I can write custom CSS locally for some sites so I can make it a bit nicer? > Is there any extension for these browsers? Oh yes too - that's absolutely necessary for todays web! Would be lost without TONS of css tweaks for lots of websites (incl. MSFN), and still growing fast. For this purpose Mozilla browsers always used this file: ../currentprofile/chrome/userContent.css By default it doesn't exist, look instead for "userContent_example.css", copy and rename it. Just a little syntax example I made for youtube: @-moz-document domain(youtube.com) { /* SHOW COMPLETE lower description text in classic view! */ div#watch-description-text { height: auto !important; max-height: none !important; } } Am sure there are also addons for easier userstyle handling. But personally I'm just manually editing an ever growing file adblock.css in my old browser, because unlike userContent.css, this one could be toggled on-off during a session. Although must sadly say, again since 1-2 years, in ancient KM1.6(FF3.5) the number of websites needing such tweaks is exploding! :-( Almost find myself fiddling and cursing daily now. And it keeps getting harder. More and more often am now forced to give up and just kill all styles completely. So far was using macros by menu on-demand only, but just yesterday was getting so fed up, started consider to do it automatically onload for some domains. This engine's CSS is just getting far too old, more and more overlapping elements in websites, more and more show wrong page width, either far too narrow or far too wide :-( If you'll manage to use Firefox9 engine or even younger, things will certainly be better. Bruninho said: > I'm almost accepting defeat and resorting to RetroZilla and Netscape 9.0.0.6 with NSS updates No, this can't be. Seriously, I can't understand at all that you shouldn't be able to move past FF2 engine in your VM, despite even most current KernelEx installed. Is this a restriction of VMs perhaps?? Too weird. My old K-Meleon1.6 with gecko 1.9.1 (as in Firefox3.5, but probably KM1.7alpha/gecko3.6 too) did run very stable with basic KernelEx452, until just 1-2 years ago. Until worldruler Google started destroying the web by enforcing TLS1.2 everywhere. Although have heard that supposedly website authors must only offer it as primary choice, and fallbacks for weaker browsers are allowed, yet most don't care and send exclusively TLS1.2 now. But despite that cipher prob, my browsers were always rather stable, did not crash hardly ever, and only showed normal page load errors for missing ciphers. Crashes happened very rarely, mostly when experimenting with something. Sadly that has changed lately, despite still using the same browser. More and more crashes now, even up to several times a day (ARRGH) No idea how many may be caused by too modern websites now (despite JS disabled), or by KernelEx16 probs, or by later added TLS1.2, or by my various browser tweaks and macros (a few do inject JS on page loads), or due to my absolutely lousy web connection, or especially my sometimes completely full disk partitions, or extreme lack of video RAM, or perhaps even by the second RAM stick added just recently, which helps much less as hoped. Am suspecting that all those points are probably contributing their share of probs. But you have a lot more resources, disk space and RAM etc., speed, and younger KernelEx with more fixes! So this really makes no sense that you should be stuck with basically the same ancient FF2 browsers running on vanilla Win98 already - this just can't be. At the very least Fx3.6 should run okay with basic KernelEx. Somewhere something is wrong... It's well possible slightly younger browsers like Fx3.6 may need SSE. Do do you have this enabled in this VM? And when using the modern KernelEx version, do you have copied over the db-files necessary for httpS sites? (from mozilla profiles created with old kernelEx452)

Hmm... perhaps there's still some hope for Opera12 with better TLS? I don't get any of this, far too complicated and technical, but perhaps anyone else: https://forums.opera.com/topic/5372/opera-12-17-no-longer-works-with-https-for-me/31 Huh?? Mozilla....? "Older versions of Opera (versions 9.5 through 12) use Opera's online root store. Until fall 2013, the online solution was built on Opera's root store program, but after that it is based on NSS by Mozilla." Some Opera setting-tips for Win98: http://kmeleonbrowser.org/forum/read.php?1,145647 Disabling OCSP allows to load more sites On the other hand it can slow down browsing, even extremely

Old Opera for vanilla Win98 and SSL, a perhaps interesting (?) discussion from 2016: https://forums.opera.com/topic/16565/what-is-causing-security-issues-can-i-fix-this If I get this right, a vanilla win98 user could access MORE sites with a certain old Opera10 build as with a newer 10.63?? > I tried most of the newer Opera version up to 10.63 and found that the same security issues apply but Opera 10, build 1750 will now load this web site Then someone remembered silent forced TLS updates to some old versions? > sometime after SSL3 was finally cited in 2014 as insecure, Opera pushed out a silent, forced 'fix' to various Olde Opera version users which forced the security protocol enablings to be set a certain way to enhance security but which avoided selection combinations that had earlier destabilized the protocol negotiations for some users.