Kiwix has curated¹ as well as complete offline versions of Wikipedia, WikiSource, Stack Overflow, Project Gutenberg (including of course the complete works of Shakespeare), Khan Academy LearnStorm, various other Khan Academy focused collections, TED Talks, CrashCourse, Super User, MathOverflow, Server Fault, lots of other Stack Exchange sites, the OpenStreetMap Wiki, ArchWiki, Ask Ubuntu, etc. You can browse through their packages at https://library.kiwix.org/ and download things like individual books from Project Gutenberg without the Kiwix application. Or you can download the files, directly or via torrent. Kiwix hides them in .local/share/kiwix.
The "standard" Kiwix app is a hypertext GUI with a built-in search engine, but you can also run it as a web server that serves up content on, for example, a local network, or just to your browser on the same machine. It's the same server running at the site above.
Kiwix is packaged in a lot of Linux distributions, including Debian (Buster, Bullseye, Bookworm, and Sid) and Ubuntu. In addition to desktop Linux, Kiwix runs on Android. Maybe some proprietary OSes too, I forget.
I don't know of an educational Linux distribution that comes with Kiwix already set up.
Speaking of OpenStreetMap, you can also download the OSM data and then use an offline map viewer like JOSM, Mapnik, or Tileserver, the latter two of which are web servers. I use OsmAnd~ on Android (available on F-Droid) but there are dozens of alternatives. I don't know which ones are best.
DebianEdu does package a bunch of education-related software, including both specifically didactic programs like Kgeography and non-didactic programs that can be useful for topics people commonly encounter in classes like Gpredict: https://blends.debian.org/edu/tasks/
If you just want a built-in dictionary to access offline, on Debian or Ubuntu, type "sudo apt install dict-gcide".
I don't think there's a free equivalent of Mathematica, but there's a lot you can do with Jupyter, Matplotlib, Numpy, SciPy, and Sympy, or with Octave, or with R; in some areas these programs exceed Mathematica's capabilities. Pari-gp is pretty great for number-theory stuff.
Sonic Pi looks pretty great, and it's free software. Popular alternatives include PureData.
More old books than Project Gutenberg and WikiSource have would be pretty great too. The Internet Archive in particular has a huge quantity of books scanned; relevant to the dictionary topic, they were kind enough to let me use their equipment to scan the first edition of the Oxford English Dictionary some years ago; each volume is a separate item of about 100 megabytes as PDF or DjVu, one example being at https://archive.org/details/oed9barch. It really rewards browsing, but looking things up in it is a bit of a pain: you need to first find the correct volume, then the correct page, then the correct column, and finally the position in the column where your desired entry starts. PDF readers are really not the ideal tools for that, particularly when they need a second or two to decompress each page image.
At any rate, you could imagine topical, selected collections of Internet Archive scans of public-domain books; for mechanical engineering, for example, you might group together a public-domain version of the Machinery's Handbook (135 MB), Reuleaux's Kinematics of Machinery (86 MB), Reuleaux's The Constructor (40 MB), Roe's English and American Tool Builders (19 MB), Brown's 507 Mechanical Movements (18 MB), and Fox's The Mechanism of Weaving (32 MB), each using the best-quality scan available (ideally one of the Internet Archive's "Scribe" scans rather than the shoddy Google Books scans that are sometimes the only ones the Archive has available). Or maybe if you knew more about mechanical engineering than I do, you would make better choices than that.
If you were going to distribute files like this to a bunch of people whose computers were going to be in close proximity, like students in a classroom, it might be a good idea to give each of them different parts of the library, instead of each getting the same standardized data set. I mean, maybe everyone's computer should have all of Shakespeare and the 2.25 GB "Best of Wikipedia", but if you distribute The Constructor to one person and The Mechanism of Weaving to another, along with some kind of file-sharing system to make local copies of files that you want to look at, then the total library available to the group can be a lot bigger than each cellphone's SD card. This is not functionality Kiwix provides at present.
______
¹ God, I hate using the word "curated" this way. It trivializes the work done by museum curators to an absurd degree.
> Speaking of OpenStreetMap, you can also download the OSM data and then use an offline map viewer like JOSM, Mapnik, or Tileserver, the latter two of which are web servers. I use OsmAnd~ on Android (available on F-Droid) but there are dozens of alternatives. I don't know which ones are best.
I also use OsmAnd~, which is great, and often find myself wondering what the best equivalent is for the Linux desktop.
Have you used any of these other solutions? I've used qmapshack along with generated garmin format images, but I find the OSM viewer ecosystem perplexing and I'm realistically unlikely to spend the required time to research and understand the various options.
Mapnik is what runs the main OSM website, so I've used that, and I'm pretty sure I've used someone's website built on Tileserver, but I can't remember the details. In both cases I have no idea how much RAM or CPU the server side requires. I have vague memories I maybe tried JOSM once.
I also find OsmAnd~ very useful but I am often frustrated with both its slowness and its interaction design. I wish I had something better.
The "standard" Kiwix app is a hypertext GUI with a built-in search engine, but you can also run it as a web server that serves up content on, for example, a local network, or just to your browser on the same machine. It's the same server running at the site above.
Kiwix is packaged in a lot of Linux distributions, including Debian (Buster, Bullseye, Bookworm, and Sid) and Ubuntu. In addition to desktop Linux, Kiwix runs on Android. Maybe some proprietary OSes too, I forget.
I don't know of an educational Linux distribution that comes with Kiwix already set up.
Speaking of OpenStreetMap, you can also download the OSM data and then use an offline map viewer like JOSM, Mapnik, or Tileserver, the latter two of which are web servers. I use OsmAnd~ on Android (available on F-Droid) but there are dozens of alternatives. I don't know which ones are best.
DebianEdu does package a bunch of education-related software, including both specifically didactic programs like Kgeography and non-didactic programs that can be useful for topics people commonly encounter in classes like Gpredict: https://blends.debian.org/edu/tasks/
If you just want a built-in dictionary to access offline, on Debian or Ubuntu, type "sudo apt install dict-gcide".
I don't think there's a free equivalent of Mathematica, but there's a lot you can do with Jupyter, Matplotlib, Numpy, SciPy, and Sympy, or with Octave, or with R; in some areas these programs exceed Mathematica's capabilities. Pari-gp is pretty great for number-theory stuff.
Sonic Pi looks pretty great, and it's free software. Popular alternatives include PureData.
More old books than Project Gutenberg and WikiSource have would be pretty great too. The Internet Archive in particular has a huge quantity of books scanned; relevant to the dictionary topic, they were kind enough to let me use their equipment to scan the first edition of the Oxford English Dictionary some years ago; each volume is a separate item of about 100 megabytes as PDF or DjVu, one example being at https://archive.org/details/oed9barch. It really rewards browsing, but looking things up in it is a bit of a pain: you need to first find the correct volume, then the correct page, then the correct column, and finally the position in the column where your desired entry starts. PDF readers are really not the ideal tools for that, particularly when they need a second or two to decompress each page image.
At any rate, you could imagine topical, selected collections of Internet Archive scans of public-domain books; for mechanical engineering, for example, you might group together a public-domain version of the Machinery's Handbook (135 MB), Reuleaux's Kinematics of Machinery (86 MB), Reuleaux's The Constructor (40 MB), Roe's English and American Tool Builders (19 MB), Brown's 507 Mechanical Movements (18 MB), and Fox's The Mechanism of Weaving (32 MB), each using the best-quality scan available (ideally one of the Internet Archive's "Scribe" scans rather than the shoddy Google Books scans that are sometimes the only ones the Archive has available). Or maybe if you knew more about mechanical engineering than I do, you would make better choices than that.
If you were going to distribute files like this to a bunch of people whose computers were going to be in close proximity, like students in a classroom, it might be a good idea to give each of them different parts of the library, instead of each getting the same standardized data set. I mean, maybe everyone's computer should have all of Shakespeare and the 2.25 GB "Best of Wikipedia", but if you distribute The Constructor to one person and The Mechanism of Weaving to another, along with some kind of file-sharing system to make local copies of files that you want to look at, then the total library available to the group can be a lot bigger than each cellphone's SD card. This is not functionality Kiwix provides at present.
______
¹ God, I hate using the word "curated" this way. It trivializes the work done by museum curators to an absurd degree.