Can we agree on not posting anything directly from twitter? It's absolutely useless without an account.
Elon removed that restriction when he first took over, then apparently reintroduced it because of machine learning scraping the site; if you believe that. It is very annoying either way, so many interesting things that belong on their own page or even a blog lost deep in multi-tweet threads.
Twitter's just completely broken for me when logged out (no account, will not get one, unless it is to build a scraper that bypasses the login wall).
In particular, some public safety / emergency response accounts post safety-critical real-time updates. Visiting their pages when not logged in provides data that is years out-of-date.
I'm hoping this causes a rapid exodus and my one use case for twitter goes away.
That's Elon's problem. Let's not make it ours.
Soon to be more annoying because Elon has started harping about putting all of twitter behind a paywall for everyone. (Obviously cash flow aint looking good) haha
A shame, this will surely tank it. I can't imagine paying unless they also introduced guaranteed privacy (like Mastodon's data permission list - empty!) on their mobile apps. Fat chance.
I'm not sure where to go as an alternative; Mastodon doesn't have the same community or blend of personalities and viewpoints as Twitter. Ah well.
Hahahahah. Quite the knee slapper.
I’m sure getting advertisers a better ROI is the goal. I’m glad someone is doing something about the bots. Jack et all didn’t give two shits and lied to investors and advertisers. This is a welcome change.
If you want a platform full of bots, build your own. Simple as.
I’d pay if the app bundled useful communication features and gave me control over my data. Also if there were no ads. Surely this would be reasonable.
Instead it looks like we will be paying AND they will be selling our data and shoving adverts in our faces willy nilly as before.
Stuff ‘dealing with bots’, this is worse.
Isn’t this the modus operandi of every company?
Example: You pay for a vehicle and they take all of your vehicle data and resell it. Uber as another, you pay for the ride yet get ads on your journey. There are many more examples.
Just be honest about your bias against X and Elon.
Not at all, my current motor doesn't take any data from me and connects to nothing. I don't use Uber. I avoid data collection when I can.
I'm not 'biased' against Mr. Musk, I think he has many impressive accomplishments under his belt. I just won't blindly praise everything he does. I wish his 'anti-establishment' senses would make him more anti-bulk collection, it seems inconsistent to me.
Looks like https://addons.mozilla.org/en-US/firefox/addon/twitter-to-ni... might be useful?
Nitter.net doesn't play videos. EDIT: This is false, I made a mistake on my side. Apologies.
Maybe a paywalled website that is actively hostile to non-users shouldn't be considered a source of news ?
> Nitter.net doesn't play videos.
I assume the new standard will be the top comment is a nitter.net link similar to how the top link on paywalled news sites is an archive.is link.
The link loads correctly for me in an incognito tab. Maybe it was something temporary when you tried? Nitter also does the job.
do you see all the tweets or just the top one?
There should be a communal throwaway account or some proxy site that uses one
Replies are retrievable via the API so any Nitter instance does the job for anonymously browsing Twitter.
But yeah, I also don't appreciate the restrictions.
Just a reminder that WINE  runs a lot of new and old Windows software pretty well, so that anyone wanting to consult the database could either load it either from a desktop Linux machine, a Linux VM, or even have it boot it directly to the database software since it can be easily embedded in a self booting Linux distro contained in a pendrive along the CD's data.
Windows 98 apparently runs fine in dosbox (dosbox-x?). My dosbox install is locked on Win 3.11, since that's a better compatibility sweet spot for me.
Anyway, that should provide much higher-fidelity emulation than Wine, and is also easier to keep off the network, etc. Also, I guess it runs on Arm processors these days (since it's an emulator, and not a hypervisor):
Wine is generally more convenient if it works though.
> it can be easily embedded in a self booting Linux distro contained in a pendrive
Very interesting. Is there a framework or helper for this?
Seems useful for things like games or audio DAWs in particular.
Not sure it does exist, but shouldn't be hard to do by hand. One could either modify a live distro to embed the necessary software, or simply run the install process with a USB key as target, then boot from it just like an internal disk and proceed to install WINE and remaining software.
Once WINE is installed, which is straightforward using the standard package managers for each distro, it runs automatically once a Windows application is clicked from a desktop windows, or called via "wine <windowsapplicationname>" from command line or a script, so that for example, if the database application in the subject is configured to auto start after the window manager is loaded, it goes straight into the Windows application as if it was part of the system.
It shouldn't be that different for DAWs.
By the way, I experimented a while ago with Alpine Linux and Yabridge to achieve a very small system that works as a host for both native and Windows .vst plugins. The goal would have been a MiniPC configured as synthesizer that boots very quick and does just that, with both studio use and live performance in mind. I got to the point plugins were converted but loading them failed. Probably using a musl based distro was asking too much, but I'll try again in the future (just moved to a new house, everything is packed) as Alpine Linux is so much faster and smaller compared to other distros that to me it makes the best candidate for building appliances where the operating system exists for the sole purpose of loading a dedicated software, therefore must not get in the way with resources consumption, automatic updates etc.
Buildroot is a good tool to build a custom Linux. Wine is available as a package.
the keyword you're looking for might be 'kiosk'.
nowadays https://porteus-kiosk.org seems to be recommended (but i haven't used it).
Wine runs 32 bits windows binaries on a 32 bits linux system and 64 bits windows binaries on a 64 bits linux system.
You can run 32 bit Windows binaries in 64 bit Linux if you install the appropriate libraries. In Debian, you can enable multiarch for this purpose.
Layers upon layers of emulators lead to lost databases.
It would be better to migrate the data to a very common format.
Wine is not an emulator.
WinE = Windows Emulator, unless someone changed the acronym since I was playing with WinE in the 1990's.
The acronym was changed to “Wine is not [an] emulator.” That’s the joke.
Agreed, but we don't know anything about the engine and formats used in that database. Given the age I would guess something from Borland or Microsoft, but could be encrypted in a way that only their software can access to it. Hard to know without examining the CD.
It’s Microsoft Access, ”works only on Windows 98” seems highly doubtful, especially as the screenshot in the post is Windows Vista…
I've never gotten familiar with the Office suite, but I ported a few graduate research databases built in Access 2003 to LibreOffice Base pretty painlessly around a decade ago.
The department discovered their newer versions of Office weren't backwards compatible, so they just kept the last PC everything worked on. It wasn't a Pentium D-series, but it was pretty close.
It's probably most likely that OP on Mysterious Twitter X simply doesn't know that's Windows Vista.
It's ancient Windows, OP doesn't know Access, UI looks like something from the 90s, Windows 98 is the most famous of the 90s Windowses(?).
As evidenced by how it's seemingly running fine on Windows Vista using whatever version of Access that is, either:
A) Any newer version of Access should be able to at least open that thing fine.
B) Whatever version of ancient Access that will open that should run just fine on Windows 11.
Microsoft's commitment to backwards compatibility shines, especially in these types of circumstances.
I think the spanner in the works is that the CD-ROM bundles an MS Access runtime executable that is 16-bit and therefore will not run on 64-bit Windows. If you look in the top left of the window in the screenshot, the icon looks like Access 2.0, which was designed for 16-bit Windows.
Given, as you say, Microsoft's commitment to backwards compatibility, it is very likely in my opinion that a modern computer running 32-bit Windows 10 (not Windows 11, as they dropped the 32-bit version and therefore finally dropped 16-bit software support entirely) would run this software fine. It is also likely, as you say, that modern Access could import the underlying database (does Access 2023/365/whatever import Access 2.0 databases? Who knows! Not I, I don't touch the stuff ;)
I think their main computer being a Mac without a CD-ROM drive may be contributing elements as well. Interesting juxtaposition that they are struggling to take the files off a cd but confident they can produce a web database. But seems they found help so nice it's all ending well.
I vaguely recall bugs where windows OS version prevented access, rather than office version.
I’m glad the access mess is over, I do wish you hadn’t triggered that memory though :P
My boss uses aceess on windows 10. It still works.
The Air Force Historical Research Agency should be contacted to figure out if it is actually important data. I would be very surprised if that old cd is the only source of this information.
edit: Reading the third paragraph here it seems that all of this data is definitely stored at different museums and libraries.
"works only on Windows 98"
Pic is of Vista running on a HP DV6000.
Perhaps Vista was the last OS to have a 98 application mode.
He could do a Vista (or earlier) VM running on Hyper-V in Win 10/11 Pro (or hacked Home). If the WWII CD is ripped to an ISO, it'll mount like a CD.
Vista install ISO: https://archive.org/details/vista_x64
Windows 98 install ISO: https://archive.org/details/windows-98-se-isofile
I don't see the discussion on Twitter as I have no account and do not see.the answers on Twitter, but don't you just need virtualbox?
CD-Roms are still there and the the underlying tech is also still the same.
Looks like it's an Access database. Perhaps convert it to SQLite and publish with something like https://datasette.io/?
I think the problem is, thread author doesn't know how to rip an ISO of the CD or move the database out; looks like they are getting help already though.
Yes, if the author was technical enough to rip an ISO and host (or seed) it somewhere, there would already be 3 Github projects linked in this thread extracting and visualising the data.
If they manage to get the ISO available can someone please share here ? I really don't want to sign up to Twitter just to follow along.
VirtualBox dropped support for 9x versions of Windows some time ago, and although it's "possible" to get it working, it works very badly.
Just use https://86box.net/ it's updated regularly and works wonders!
How is it possible to do this? Isn't all you need is to emulate is at least a 386 or 486? They removed emulation of that cpu target?????
Not just the CPU, but also the older chipsets that Windows 9x thinks is current. It only supports emulation of chipsets much newer than those versions of Windows.
Where did you get that idea? Windows 98 is still a default template in VirtualBox 7.0 and fully supported:
You just don't get accelerated graphics or USB sharing.
You're being very generous. With no guest additions, I'd say that's "unsupported". It will work as well as any other arbitrary Intel software, but it won't be usable.
I'm not being "generous". If anything, I just think you're using an overly ambitious definition of "supported".
Oracle isn't going to guarantee your Windows 7 install works or help you debug issues. And running an OS non-accelerated is a perfectly valid and functional way of doing so, even if it doesn't fit your personal metric of usability.
Regardless, you implied that something changed for the worst. VBox offers as much "support" in version 7.0 as they did in 1.0. If anything it's better today as VBox used to be unbearably unstable in VESA or SVGA modes:
And running 16-bit applications (WoW/ntvdm) on 64-bit VBox installs used to crash the whole machine (or run unbearably slow) due to incomplete CPU emulation, something that has since been remedied. A feature that only old 32-bit Windows OSes use.
It happens to work but is not really supported. Mouse is super laggy, there are no guest additions and last time I checked it used a full CPU core at all times.
> You just don't get accelerated graphics or USB sharing.
Every OS that has to use VESA in a VM is slow. It doesn't mean they're "unsupported". Unsupported would imply that the OS doesn't boot/function properly or that you have to do abnormal things to get it to work.
It's not like Oracle offers some sort of guarantee that your Windows 7 guest will function or offer support if you have issues.
> Every OS that has to use VESA in a VM is slow.
Actually, that is not necessarily true. If properly done, VESA is just a large shared memory framebuffer which is practically as fast as you can get on a VM, barring 3D acceleration which the GUI of 9x doesn't really benefit from.
Heck, for a fun experiment, Haiku on _real hardware_ using "VESA" (actually, the EFI GOP) beats the shit out of Windows 11 with accelerated GPU drivers. Even dragging windows around feels much faster in Haiku. Obviously 3D is going to be a problem, but...
Sorry, I said "a VM" when I should have said "VirtualBox". VBox just has a suboptimal VESA implementation. And I overloaded non-accelerated into that for simplicity's sake.
Qemu shows that VESA/SVGA drivers can perform decently, the VBox team just hasn't prioritized making them do so. But the point stands that a non-accelerated video option will perform worse than an accelerated one for all VMs for anything but the simplest of blitting pipelines.
What do you mean? The VBox VESA BIOS is buggy? (There's not much else in a "VESA implementation") I really don't think that's the case. It is a bit slower than VMware, but never felt ridiculously so. And definitely qemu feels slower, and IMHO has a much worse and less compatible implementation (yes, the VBox BIOS forks from qemu, but at a point where qemu cared a bit more about retro-compatibility than nowadays). At worst I'd argue they are practically the same implementation.
> But the point stands that a non-accelerated video option will perform worse than an accelerated one for all VMs for anything but the simplest of blitting pipelines.
That's why I put the example that Haiku on a completely not accelerated framebuffer performs miles better than Windows 11 on an accelerated 2 year old GPU. The GUI parts only, of course. Try it. It's just insane difference. And yes I do have 5k screens..
For 9x, even if it actually supported 2D acceleration, newer cards seem to do a terrible job at it.
You are exaggerating a lot. Windows 9x works perfectly fine, and is much faster than on emulators. You can even use USB tablet emulation to get as good as it gets mouse support.
What does "supported" mean? Are you a paying Oracle customer? Because otherwise _nothing_ is supported.
If you pay them, they for example support several 9x era hardware (like the SB16).
VirtualBox defaults to emulating a PIIX3 chipset which is absolutely Windows 95 vintage.
It actually works very well, and there was an HN top post a couple weeks ago about how someone had even written a 3D capable video driver for 98 and VirtualBox.
I agree but you don't need to run Windows 9x. NT 4 or 2000 should work fine and are well supported in VirtualBox.
Glad to hear people have reached out to help, but the sad thing is for every project like this there are a hundred more that die a quiet death every year. There's a universe of important and unique data tied to history and genealogy blogs, web databases, 20th-century software programs, and hard drive archives maintained by a single person.
One of my favorites: https://fultonhistory.com/ (click on the "Go and Search My Archive" link). Tom has done an amazing job of scanning and posting old historical photos and newspaper archives, but will it last once he's no longer able to maintain it?
I often hear suggestions like "let Ancestry take them over" or "the Internet Archive is the solution." I have to ask: Will either be operational in 20 years, or the data on them as easily accessible as they are now?
Ancestry is currently operated by Blackstone, the third or fourth PE firm to do so in the past 15 years. Blackstone has no qualms with deleting services that no longer meet its needs (https://scottishgenes.blogspot.com/2023/08/ancestry-disconti...) or paywalling records and then jacking up the price every few years (https://www.ancestry.com/corporate/blog/were-increasing-our-...).
IA's web archive is a great resource, but it always seems to be begging for money and its founder's futile campaign against book publishers isn't helping.
Fulton History is particularly frustrating as it’s a passion project and amazing resource, but the owner is a little… mercurial.
It will absolutely die with him. Its unfortunate that when the NYS Legislature appropriated money to scan thousands (millions) of newspapers to microfiche, they did not provide any funding for digitization.
> Glad to hear people have reached out to help, but the sad thing is for every project like this there are a hundred more that die a quiet death every year.
CDROM is a terrible way to store data. Any frontend to dd could back the CDROM up 1:1. That should be the first and primary focus. After that and an upload of however long (depending on internet connection), the wisdom of the crowd could be utilized. Even if the requirement of Windows 98 is correct, any VM (perhaps even ReactOS) could access it.
IIRC, there used to be a dd-like tool, maybe cdrescue? (a Linux command-line tool), that would recheck the data extracted, and then do a "best of n" check on any spurious sectors of the disc.
It's been at least 10 years since I had a CD/DVD drive though, so it might have been a mode of one of the dd_rescue/ddrescue/dd-rescue programs??
Probably just ddrescue by GNU, which can also be used on CD drives.
For audio CDs, I've always used EAC  with secure mode. I believe I even used it within Wine. It yielded better results than cdrecord.
Morgoth, FlashFXP, Preee, hkSFV, DCPP, and all that beautiful Windows software also worked in Wine.
> I often hear suggestions like "let Ancestry take them over" or "the Internet Archive is the solution." I have to ask: Will either be operational in 20 years, or the data on them as easily accessible as they are now?
And on the pedestal, these words appear:
My name is Ozymandias, King of Kings;
Look on my Works, ye Mighty, and despair!
Nothing beside remains. Round the decay
Of that colossal Wreck, boundless and bare
The lone and level sands stretch far away.
I think any resource of data like this - assuming it's a noncommercial project - should have an export functionality available at all times, and an ask to any visitor to download the file, archive it, and clearly label it.
I like to believe that the best data repositories are the ones that are forgotten.
That said, I hope someone comes up with a consumer-affordable, long-term storage medium soon.
It a lesson in how important it is to choose the most open/free and widely supported formats when creating information you want to share with the world.
I find myself thinking about the conundrum of long-term digital archiving pretty often. It's a very interesting question.
I think the first biggie is non decaying storage. Outside of some paper, everything seems to decay fast, and even if stored correctly. Film goes bad in climate controlled vaults at times.
Those DVDs made of rock supposedly last a long time, but it would be nice to have something that lasts 1000s of years.
If you're talking about analog archival that is a whole other ball of yarn, but in terms of digital then I would think that the thing to do is use only lossless compression, bit-level fidelity error correction, and multiply-redundant, multi-region storage. You must plan on storage media, providers, and entire networks partitioning, disappearing, and failing.
You have to actively have an archive, with archivists. The geek stuff is secondary.
Something that can only survive if it has long-term funding is not really a solution to long-term archival.
Why not to donate this data to a nearest library?
1) Boot Access inside Wine and use it to export the tables into csv or similar. From there it's a relative doddle to import them into SQLite, Postgres, etc.
2) Use one of the open source Access file readers (e.g., Jackcess) to do the same. Make a backup first!
It looks like an application based on Microsoft Access.
Where I work we still occasionally use Office 97 with all the strange things it does to take over your desktop and that paperclip. It works just fine in Windows 10 and could probably open that database and see the SQL tables, it would probably not be hard at all to load the SQL tables in Postgres. I did a data import project for that kind of database and I am waiting for the QA people to tell me I did it right.
I do know I was unable to open the database w/ the current Office 365 and was told to go back and get the old Office.
One option: Photographs or Screenshots -> Tesseract OCR -> ChatGPT cleanup -> CSV -> Human validation -> public GitHub repo
My (late) uncle Ralph is probably in that database. Can I get a copy of the ISO? I'll help in any way that I can.
I think you may wanna reply on Twitter.
I don't have an account there and I am not going to create one
This looks like the perfect usecase to extract the data to sqlite, and slap a quick SQLPage UI on it !