Ok, I have now become officially obsessed with my dumb Dos-Unix toy
So I'm going through and "Filling in the gaps" to flesh out the functionality of this idiotic adventure. I've now resigned myself to create the tools that are missing.
For example, you know that Linux has that my little DOS-Unix thing doesn't?
A motd
Now, you may think, "Really? motd? That's what's bugging you?"
Well, no, it's just that most of my "Unix" side environment is being set up by /etc/profile before bash runs. I can shove a motd in there, but why when that is actually handled by login. However, DOS (nor any of the DJGPP gnu utilities) come with login. Why? It's a single user system with no concept of ownership anyway
However, it would be nice to have home directories, and I get an motd, and for funzies when you log in, you get your own username... even though you are still UID 0.
Problem: DGJPP doesn't come with crypt()
Yea, I know, I know, why bother when anyone can just go to /etc/passwd and just edit the file... Well, I don't like seeing passwords in plain text and it irks me to see them. I don't need shadow passwords anyway so I guess I can use a simple md5 hash . With no crypt() I'll have to find some md5 digest code and see if I can get it running under DOS in DJGPP. So now I have to write my own "OMG never use this in real life" crypt() function so I can implement /etc/passwd so that I can then implement login so I can get a motd and user directories... in DOS...
The only nice thing about a Windows laptop over a Mac laptop in my experience is that Windows Outlook is still better than Mac Outlook.
Other than that I like having native SSH that "just works". Maybe the WSL helps with that though, but I like being able to just fire up iTerm and go to town.
Back in the day there were crucial hardware differences: 10 years ago if I had $2000 to spend on a laptop I'd get a Mac every time. I'm going to guess that's less true now.
i've been using mail for mac and it can't be better than outlook..... I'm pretty sure that I've hit some sort of untheorized performance case in Mail with an unpruned 20 year old gmail inbox sometimes I open the client and it just cannot get email.... in ANY of my inboxes. i just get an empty message list. and I wont get one until i reboot.
that said I've been using macs since the heydey of the iPod and while the hardware has grown infinitely more impressive, the software I am afraid is in linear decline. the software performance of my Macbook Pro are.... worrisome... and I really wish I could get away with a PC environment right now... but I've tried porting my docker stack to PC and the results are shockingly inconsistent and bad... what a shame
The job adverts at the side of Stack Overflow are just about perfectly placed. "Tired of beating your head against whatever fucking problem brought you here? Fuck those guys and come work on some other fucking problems!"
The only nice thing about a Windows laptop over a Mac laptop in my experience is that Windows Outlook is still better than Mac Outlook.
Other than that I like having native SSH that "just works". Maybe the WSL helps with that though, but I like being able to just fire up iTerm and go to town.
Back in the day there were crucial hardware differences: 10 years ago if I had $2000 to spend on a laptop I'd get a Mac every time. I'm going to guess that's less true now.
i've been using mail for mac and it can't be better than outlook..... I'm pretty sure that I've hit some sort of untheorized performance case in Mail with an unpruned 20 year old gmail inbox sometimes I open the client and it just cannot get email.... in ANY of my inboxes. i just get an empty message list. and I wont get one until i reboot.
that said I've been using macs since the heydey of the iPod and while the hardware has grown infinitely more impressive, the software I am afraid is in linear decline. the software performance of my Macbook Pro are.... worrisome... and I really wish I could get away with a PC environment right now... but I've tried porting my docker stack to PC and the results are shockingly inconsistent and bad... what a shame
Yeah, I suspect Docker is in that category of "originally Linux-based tool that is a hell of a lot easier to get working properly on a Mac than Windows".
I tried using Apple Mail for 5 minutes when I first got a Mac, discovered there is no keyboard shortcut for "next unread message" and immediately stopped using it. I still use Thunderbird for non-work things.
this made me realize that even though I can't run docker on windows... there's no reason I cant just put a virtualbox on and run it that way... with my water cooled ryzen threadripper... yeah.. the performance is *quite good*
so now my laptop can be reserved for laptoping instead of idling my local dev environment all day long
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
Docker for Desktop on Windows is actually very simple to setup. It will use WSL2 as a back end out of the box at this point. For most users it should be completely painless to get Docker going on Windows.
So I'm getting intensely bored at work doing some pretty straightforward SQL stuff. I've been doing "data" (SQL/Making graphs/getting information from here to there) work for a while but I'm looking to expand my skillset to general backend stuff. I want to build a skillset that means I can avoid ever needing to make graphs as part of my job, because that was boring.
Right now I've been doing a bunch of Pluralsight Python course stuff because I find it a fun language to work in and I'd like a more comprehensive grounding. Anyone have suggestions for interesting places to build some new skills?
The job adverts at the side of Stack Overflow are just about perfectly placed. "Tired of beating your head against whatever fucking problem brought you here? Fuck those guys and come work on some other fucking problems!"
I once got my partner and several other folks a raise because their company advertised on stack overflow for more money than the company paid current employees. They all went to talk to the boss and were like "so why would a brand-new person with our qualifications be worth $X?" and they got bumped up the next paycheck.
So I'm getting intensely bored at work doing some pretty straightforward SQL stuff. I've been doing "data" (SQL/Making graphs/getting information from here to there) work for a while but I'm looking to expand my skillset to general backend stuff. I want to build a skillset that means I can avoid ever needing to make graphs as part of my job, because that was boring.
Right now I've been doing a bunch of Pluralsight Python course stuff because I find it a fun language to work in and I'd like a more comprehensive grounding. Anyone have suggestions for interesting places to build some new skills?
Disclaimer: I am not at all an expert in this field. The following advice is based on my personal experience in the very recent past, as explained below.
Just this past week I've been doing some prep work for an interview as a backend developer, coming from an algorithm engineer background. I had no previous experience with anything backend, aside from some occasional cooperation with the backend and devops engineers in my previous job. (I passed the interview woooo)
So, while this is not quite what you're asking for, my suggestions is: Build and set up a small web app on one of the cloud services. I'd recommend GCP, which give you 300$ to burn over the course of a year (plus there are always-free usage quotas, which you're unlikely to exceed with a pet project), but the other big ones are just as valid. Get some data in a database, build an OpenAPI (or maybe GraphQL) API, set up a rudimentary frontend. This approach won't lead to deep expertise in any particular field, but it will give you a good grasp of all the different parts of the system. In my experience at least, it was a good way to start.
Also re: python - it's been many years, but I remember the Codecademy python course being pretty good.
Oh, and absolutely learn docker. Possibly the single most important piece of technology when it comes to backend work these days, imho.
Give me a kiss to build a dream on; And my imagination will thrive upon that kiss; Sweetheart, I ask no more than this; A kiss to build a dream on
Docker for Desktop on Windows is actually very simple to setup. It will use WSL2 as a back end out of the box at this point. For most users it should be completely painless to get Docker going on Windows.
Beware the auto-updater though. It has broken my installation no fewer than three times over the past two years.
Anyone have any idea why a AWS Aurora database would only be using 1 core no matter how many connections I open to it to INSERT data.
This happens whether I'm writing to one table or multiple tables (I thought initially it might be some kind of write lock but the multiple table scenario eliminates that) from a single machine.
Does Postgres put all connections from a single machine on the same process?
way late to the party here but no postgres doesn't just generally do that... I run a multi-queue data import script against my database and it always opens up multiple processes and I get really good utilization out of my 16 core VM...
that said... it's an AWS product so who fucking knows what they do to the configs...
I'm getting suspicious of the official postgres JDBC driver.
I went and opened up two Datagrip consoles side by side and ran parallel queries and they, as expected, each got their own core and I pegged out 2 of the 4 cores at 100% each.
So this suggests that in some way when I think I have requested two separate Connection the Driver is actually multiplexing a single connection. This would be a huge surprise because I read the code and didn't see anything like that going on but it is starting to look like the only thing it could possibly be.
Docker for Desktop on Windows is actually very simple to setup. It will use WSL2 as a back end out of the box at this point. For most users it should be completely painless to get Docker going on Windows.
Beware the auto-updater though. It has broken my installation no fewer than three times over the past two years.
They have this new thing where there's a "Skip update" button in the update notification, but it's a premium feature. It's amazing!
So I'm getting intensely bored at work doing some pretty straightforward SQL stuff. I've been doing "data" (SQL/Making graphs/getting information from here to there) work for a while but I'm looking to expand my skillset to general backend stuff. I want to build a skillset that means I can avoid ever needing to make graphs as part of my job, because that was boring.
Right now I've been doing a bunch of Pluralsight Python course stuff because I find it a fun language to work in and I'd like a more comprehensive grounding. Anyone have suggestions for interesting places to build some new skills?
The job adverts at the side of Stack Overflow are just about perfectly placed. "Tired of beating your head against whatever fucking problem brought you here? Fuck those guys and come work on some other fucking problems!"
I once got my partner and several other folks a raise because their company advertised on stack overflow for more money than the company paid current employees. They all went to talk to the boss and were like "so why would a brand-new person with our qualifications be worth $X?" and they got bumped up the next paycheck.
If you want to stay in the data space, I'd say learn the key concepts behind MapReduce (how much you play with hadoop/spark is up to you, it's the conceptual framework that's important for scaling data operations infinitely in the cloud). Really learn permissions and service interactions if you want to infra stuff, as a lot of it is navigating how to make infrastructure as code and not introduce massive security holes. Netflix tech blog has a lot of good conceptual and practical stuff to understand and try.
Docker for Desktop on Windows is actually very simple to setup. It will use WSL2 as a back end out of the box at this point. For most users it should be completely painless to get Docker going on Windows.
Beware the auto-updater though. It has broken my installation no fewer than three times over the past two years.
They have this new thing where there's a "Skip update" button in the update notification, but it's a premium feature. It's amazing!
I guess another win for Linux. To be a little fair, according to this (https://www.docker.com/blog/changing-how-updates-work-with-docker-desktop-3-3/) you don't have to install, but I guess you have to keep pressing "SNOOZE": "Free users will not be required to install any particular update, but they will continue to get reminders to update."
I suppose they are desperate for ways to monetize Docker.
for a while though the snooze button didn't exist, but I am relieved it's been added... well... i guess it doesn't matter to me since I have other legitimate reasons to pay for docker pro but if I didn't...! then i'd be safe from updates
that said, after a rocky summer, the last two engine updates did not tank my local environment, finally
We have a code quality step in our CI that runs linters, and if they have issues, you need to look at the output of that pipeline job. Not a big deal, but a tiny annoyance.
Today I stumbled across Reviewdog, and tl;dr whipped up a PoC for our GitLab pipelines:
That's just piping our regular linter through Reviewdog, and I like it a lot already.
We have a code quality step in our CI that runs linters, and if they have issues, you need to look at the output of that pipeline job. Not a big deal, but a tiny annoyance.
Today I stumbled across Reviewdog, and tl;dr whipped up a PoC for our GitLab pipelines:
That's just piping our regular linter through Reviewdog, and I like it a lot already.
Doesn't GoLand already pick up unused code? At least that's been my experience.
Doesn't GoLand already pick up unused code? At least that's been my experience.
It certainly has the capability, and there's different syntax highlighting for it, yeah. The "Look for issues" thingie might also show it, but then you'd have to actually run it.
It's also nice only having it as part of your PR/CI pipeline. I don't give a monkeys what's going on while you're developing, but as soon as its in PR the less ways there are for bad code to get merged the better.
It's also nice only having it as part of your PR/CI pipeline. I don't give a monkeys what's going on while you're developing, but as soon as its in PR the less ways there are for bad code to get merged the better.
I'm an unexpected fan of having a local linter yell at you, but for some specific errors, actually fail your local builds. I'd have expected to hate it, but like... It's easier than going back and fixing it later, you know? (Don't ask me which in particular, I can never even remember where we keep the config...)
All right, I finished porting my mathemodeon to C#, now to code a GUI.
Does anyone have experience with WPF? I'd like to know if I can do certain things.
I haven't worked with WPF professionally for a couple of years, and that was pretty basic data-entry systems - but I'm happy to try and answer any questions you've got. Generally you can do pretty much anything you can do in any other GUI framework - it's just a question of how easy/difficult it makes it...
If you're looking for a framework, I recommend Stylet. If fixes most of the issues I had with Caliburn.Micro whilst being a lot easier to pick up. I've been porting all my personal projects over to it, and its so much nicer to use than the other WPF frameworks i've looked at.
It's also nice only having it as part of your PR/CI pipeline. I don't give a monkeys what's going on while you're developing, but as soon as its in PR the less ways there are for bad code to get merged the better.
Yeah, we used to do the whole test/build pipeline on every push, but we recently switched it to run on merge requests only to save some resources.
It's also nice only having it as part of your PR/CI pipeline. I don't give a monkeys what's going on while you're developing, but as soon as its in PR the less ways there are for bad code to get merged the better.
Yeah, we used to do the whole test/build pipeline on every push, but we recently switched it to run on merge requests only to save some resources.
Our front end Jest tests are so very thoroughly fucked. Currently they take about 14 minutes to run in their entirety.
Working with Microsoft Access is seriously the most miserable experience you will ever have.
Nothing is easy, nothing makes sense, and everything breaks all the time for no reason.
The only thing that even comes close is SharePoint.
We had a big push into SharePoint back in 2012-2013 when I started my current job. Lots of customization requests and complicated approval flows and things like that fell onto our group and it was a nightmare.
Fortunately the main proponents retired or left so it's still limping along only being used as a glorified shared drive.
Working with Microsoft Access is seriously the most miserable experience you will ever have.
Nothing is easy, nothing makes sense, and everything breaks all the time for no reason.
The only thing that even comes close is SharePoint.
We had a big push into SharePoint back in 2012-2013 when I started my current job. Lots of customization requests and complicated approval flows and things like that fell onto our group and it was a nightmare.
Fortunately the main proponents retired or left so it's still limping along only being used as a glorified shared drive.
We have an Access database that's accessed through sharepoint to for data entry into a SQL Server database. It was set up years ago when we were much, much smaller than we are now. Every opportunity I get I suggest we replace it.
At my last job every department had their own homegrown ticketing system in Sharepoint (each department set up their own). At least once of those had to be purged of anything older than 6 months because Sharepoint couldn't handle the volume.
0
Options
BlackDragon480Bluster KerfuffleMaster of Windy ImportRegistered Userregular
The junior devs were eying a 3 1/2" floppy like it was a relic from a bygone age.
I suddenly felt ancient.
I've got a bunch of Infocom and Origin games from the mid80s that are on 5 1/4" floppies. I think I have some ancient 8" disks floating around somewhere as well.
No matter where you go...there you are. ~ Buckaroo Banzai
Never actually used 8" discs, but I've seen a couple around in their natural habitat.
Started with 5 1/4" floppies on the C64.
I've seen the 8" discs a few times, but our first PC had 5 1/4" floppies.
We (or more accurately my grandpa) had a tape and cartridge drive for his C64. The tape drive I think was the same form factor as regular audio cassettes.
That 286 my dad built with the 5 1/4" floppies he was always talking about buying a hard drive, which for the longest time I thought was the 3.5" disks. I was super pumped about that because that's what all my cousin's games were on and I knew I could copy them. I was extremely disappointed when that turned out to not be the case (he never did get the HDD working).
0
Options
Ear3nd1lEärendil the Mariner, father of ElrondRegistered Userregular
I started a job back in 2008 and found out that the ninnies who preceded me set up a bunch of sites using Access as the back-end database. Apparently, they couldn't figure out why the sites crashed all the time.
0
Options
OrcaAlso known as EspressosaurusWrexRegistered Userregular
I threw out my last floppy disks circa 2008 maybe when I cleaned house and realized I didn't even own a floppy drive of either variety at that point.
I should have held onto one of my old AOL disks I repurposed to a boot disk for old time's sake.
I found my 3.5" USB drive, still have that kicking.
Also found an old 5.25" disk case that had a bunch of nostalgic shit, including a few disks labeled QuickBasic, QuickBasic Help Files, QuickBasic Programs 1, etc.
Reminded me that to get the full help content when pressing F1 on a function, I'd have to swap floppies.
I bought two USB floppy disk drives a few years ago because I hadn't backed up my Wolf3D Episodes 1-6 disk (it still worked!). Just FYI, even though they are under different brand names, it's the same 1 Chinese company making the USB floppy drives (the first one I got wouldn't read disks). As long as the one I have now never craps out, I can load the very rare floppy disk that comes along. I did keep an internal floppy drive from the computer I built in 2004, but I would need to get some sort of converter that goes from the old floppy interface to SATA or something.
I also got some floppy disks from a friend a few years ago who happened to have some left over to help revive my old Macintosh LC from 1990. That was an interesting project.
The only nice thing about a Windows laptop over a Mac laptop in my experience is that Windows Outlook is still better than Mac Outlook.
Other than that I like having native SSH that "just works". Maybe the WSL helps with that though, but I like being able to just fire up iTerm and go to town.
Back in the day there were crucial hardware differences: 10 years ago if I had $2000 to spend on a laptop I'd get a Mac every time. I'm going to guess that's less true now.
Win10 has a native OpenSSH server and client in cmd and powershell. Also stuff like MobaXTerm if you want to get fancy, or VSCode's remote-ssh extension is a godsend for remote dev.
Posts
So I'm going through and "Filling in the gaps" to flesh out the functionality of this idiotic adventure. I've now resigned myself to create the tools that are missing.
For example, you know that Linux has that my little DOS-Unix thing doesn't?
A motd
Now, you may think, "Really? motd? That's what's bugging you?"
Well, no, it's just that most of my "Unix" side environment is being set up by /etc/profile before bash runs. I can shove a motd in there, but why when that is actually handled by login. However, DOS (nor any of the DJGPP gnu utilities) come with login. Why? It's a single user system with no concept of ownership anyway
However, it would be nice to have home directories, and I get an motd, and for funzies when you log in, you get your own username... even though you are still UID 0.
Problem: DGJPP doesn't come with crypt()
Yea, I know, I know, why bother when anyone can just go to /etc/passwd and just edit the file... Well, I don't like seeing passwords in plain text and it irks me to see them. I don't need shadow passwords anyway so I guess I can use a simple md5 hash . With no crypt() I'll have to find some md5 digest code and see if I can get it running under DOS in DJGPP. So now I have to write my own "OMG never use this in real life" crypt() function so I can implement /etc/passwd so that I can then implement login so I can get a motd and user directories... in DOS...
This is insane.
It did it again, but this time it suspected it was Japanese.
i've been using mail for mac and it can't be better than outlook..... I'm pretty sure that I've hit some sort of untheorized performance case in Mail with an unpruned 20 year old gmail inbox sometimes I open the client and it just cannot get email.... in ANY of my inboxes. i just get an empty message list. and I wont get one until i reboot.
that said I've been using macs since the heydey of the iPod and while the hardware has grown infinitely more impressive, the software I am afraid is in linear decline. the software performance of my Macbook Pro are.... worrisome... and I really wish I could get away with a PC environment right now... but I've tried porting my docker stack to PC and the results are shockingly inconsistent and bad... what a shame
Yeah, I suspect Docker is in that category of "originally Linux-based tool that is a hell of a lot easier to get working properly on a Mac than Windows".
I tried using Apple Mail for 5 minutes when I first got a Mac, discovered there is no keyboard shortcut for "next unread message" and immediately stopped using it. I still use Thunderbird for non-work things.
so now my laptop can be reserved for laptoping instead of idling my local dev environment all day long
Right now I've been doing a bunch of Pluralsight Python course stuff because I find it a fun language to work in and I'd like a more comprehensive grounding. Anyone have suggestions for interesting places to build some new skills?
I once got my partner and several other folks a raise because their company advertised on stack overflow for more money than the company paid current employees. They all went to talk to the boss and were like "so why would a brand-new person with our qualifications be worth $X?" and they got bumped up the next paycheck.
Disclaimer: I am not at all an expert in this field. The following advice is based on my personal experience in the very recent past, as explained below.
Just this past week I've been doing some prep work for an interview as a backend developer, coming from an algorithm engineer background. I had no previous experience with anything backend, aside from some occasional cooperation with the backend and devops engineers in my previous job. (I passed the interview woooo)
So, while this is not quite what you're asking for, my suggestions is: Build and set up a small web app on one of the cloud services. I'd recommend GCP, which give you 300$ to burn over the course of a year (plus there are always-free usage quotas, which you're unlikely to exceed with a pet project), but the other big ones are just as valid. Get some data in a database, build an OpenAPI (or maybe GraphQL) API, set up a rudimentary frontend. This approach won't lead to deep expertise in any particular field, but it will give you a good grasp of all the different parts of the system. In my experience at least, it was a good way to start.
Also re: python - it's been many years, but I remember the Codecademy python course being pretty good.
Oh, and absolutely learn docker. Possibly the single most important piece of technology when it comes to backend work these days, imho.
Beware the auto-updater though. It has broken my installation no fewer than three times over the past two years.
I'm getting suspicious of the official postgres JDBC driver.
I went and opened up two Datagrip consoles side by side and ran parallel queries and they, as expected, each got their own core and I pegged out 2 of the 4 cores at 100% each.
So this suggests that in some way when I think I have requested two separate Connection the Driver is actually multiplexing a single connection. This would be a huge surprise because I read the code and didn't see anything like that going on but it is starting to look like the only thing it could possibly be.
I made a game, it has penguins in it. It's pay what you like on Gumroad.
Currently Ebaying Nothing at all but I might do in the future.
They have this new thing where there's a "Skip update" button in the update notification, but it's a premium feature. It's amazing!
If you want to stay in the data space, I'd say learn the key concepts behind MapReduce (how much you play with hadoop/spark is up to you, it's the conceptual framework that's important for scaling data operations infinitely in the cloud). Really learn permissions and service interactions if you want to infra stuff, as a lot of it is navigating how to make infrastructure as code and not introduce massive security holes. Netflix tech blog has a lot of good conceptual and practical stuff to understand and try.
I guess another win for Linux. To be a little fair, according to this (https://www.docker.com/blog/changing-how-updates-work-with-docker-desktop-3-3/) you don't have to install, but I guess you have to keep pressing "SNOOZE": "Free users will not be required to install any particular update, but they will continue to get reminders to update."
I suppose they are desperate for ways to monetize Docker.
that said, after a rocky summer, the last two engine updates did not tank my local environment, finally
Today I stumbled across Reviewdog, and tl;dr whipped up a PoC for our GitLab pipelines:
That's just piping our regular linter through Reviewdog, and I like it a lot already.
Doesn't GoLand already pick up unused code? At least that's been my experience.
Does anyone have experience with WPF? I'd like to know if I can do certain things.
It certainly has the capability, and there's different syntax highlighting for it, yeah. The "Look for issues" thingie might also show it, but then you'd have to actually run it.
http://steamcommunity.com/id/pablocampy
I'm an unexpected fan of having a local linter yell at you, but for some specific errors, actually fail your local builds. I'd have expected to hate it, but like... It's easier than going back and fixing it later, you know? (Don't ask me which in particular, I can never even remember where we keep the config...)
I haven't worked with WPF professionally for a couple of years, and that was pretty basic data-entry systems - but I'm happy to try and answer any questions you've got. Generally you can do pretty much anything you can do in any other GUI framework - it's just a question of how easy/difficult it makes it...
If you're looking for a framework, I recommend Stylet. If fixes most of the issues I had with Caliburn.Micro whilst being a lot easier to pick up. I've been porting all my personal projects over to it, and its so much nicer to use than the other WPF frameworks i've looked at.
Yeah, we used to do the whole test/build pipeline on every push, but we recently switched it to run on merge requests only to save some resources.
Our front end Jest tests are so very thoroughly fucked. Currently they take about 14 minutes to run in their entirety.
I don't think we are Reacting right
http://steamcommunity.com/id/pablocampy
Nothing is easy, nothing makes sense, and everything breaks all the time for no reason.
The only thing that even comes close is SharePoint.
We had a big push into SharePoint back in 2012-2013 when I started my current job. Lots of customization requests and complicated approval flows and things like that fell onto our group and it was a nightmare.
Fortunately the main proponents retired or left so it's still limping along only being used as a glorified shared drive.
I suddenly felt ancient.
I mean
it is a relic from a bygone age
as is the zip drive, the jazz drive, and these days, the CD and DVD
I'm probably the only idiot still buying either of the last two because I don't like renting
We have an Access database that's accessed through sharepoint to for data entry into a SQL Server database. It was set up years ago when we were much, much smaller than we are now. Every opportunity I get I suggest we replace it.
At my last job every department had their own homegrown ticketing system in Sharepoint (each department set up their own). At least once of those had to be purged of anything older than 6 months because Sharepoint couldn't handle the volume.
I've got a bunch of Infocom and Origin games from the mid80s that are on 5 1/4" floppies. I think I have some ancient 8" disks floating around somewhere as well.
~ Buckaroo Banzai
Started with 5 1/4" floppies on the C64.
I've seen the 8" discs a few times, but our first PC had 5 1/4" floppies.
We (or more accurately my grandpa) had a tape and cartridge drive for his C64. The tape drive I think was the same form factor as regular audio cassettes.
That 286 my dad built with the 5 1/4" floppies he was always talking about buying a hard drive, which for the longest time I thought was the 3.5" disks. I was super pumped about that because that's what all my cousin's games were on and I knew I could copy them. I was extremely disappointed when that turned out to not be the case (he never did get the HDD working).
I should have held onto one of my old AOL disks I repurposed to a boot disk for old time's sake.
Also found an old 5.25" disk case that had a bunch of nostalgic shit, including a few disks labeled QuickBasic, QuickBasic Help Files, QuickBasic Programs 1, etc.
Reminded me that to get the full help content when pressing F1 on a function, I'd have to swap floppies.
I also got some floppy disks from a friend a few years ago who happened to have some left over to help revive my old Macintosh LC from 1990. That was an interesting project.
Win10 has a native OpenSSH server and client in cmd and powershell. Also stuff like MobaXTerm if you want to get fancy, or VSCode's remote-ssh extension is a godsend for remote dev.