14
votes
What programming/technical projects have you been working on?
This is a recurring post to discuss programming or other technical projects that we've been working on. Tell us about one of your recent projects, either at work or personal projects. What's interesting about it? Are you having trouble with anything?
I'm working on setting up an overly complex method of watching my 3D printer's progress through my media platform of choice (Emby).
The webcam stream from my 3D printer uses MJPG, which is great for viewing from a web browser, but not very friendly for many other video decoders, etc. Emby does not support MJPG for streaming purposes (Live TV, or even plugins like IPTV), so a direct stream is out.
There's a restreaming software called MediaMTX, but that also doesn't work (directly) with MJPG.
So what I've decided to do is use Neko, a VNC-esque software with broadcast capabilities, to show the Fluidd webpage from my 3D printer (which will also include current progress, temps, etc.) and convert it to an RTMP stream. I will then use MediaMTX to parse that stream into an Emby-compatible format (TBD), and access it via an M3U file published by MediaMTX. This will create a "Live TV" channel for my 3D printer!
If anyone has any other suggestions, I'm all ears, but this is the way I've come up with to show a webpage as a video stream.
i have been thinking about setting up a camera for my printer (it doesn't have one itself) manually using a usb webcam, a raspberry pi, and some software to turn the webcam into an video stream on my network, where i could view it through VLC or maybe some web interface
haven't done so yet, though. but i think it will be fairly simple and using my own camera and server means more control over the video
though it also means it's completely passive unless i do some fancy stuff with the rpi and the printer firmware and all that. but i don't really need that, just want a way to monitor from other rooms
yeah it's tough that your printer's camera forces that format which other things aren't playing well with... nice job working around the problem though!
You might look into octoprint. I ran it with a raspberry pi 3 and a pi cam for years on my Prusa MK3. It streams the camera and gives you control.
I set mine up with HA proxy, Lets Encrypt certificates, and http auth, and with those extra layers, I felt comfortable exposing it to the Internet. It has been super useful to be able to monitor a long print when I'm away and stop it if I get a bad failure before I get the blob of death. I really liked being able to preheat the bed or the nozzle before going down to change the filament, although the hot ends are so fast these days it barely matters anymore
These days I just use Prusa Connect since it's securely hosted and integrated with my mk4 and prusa one. They have the buddy cam which is also integrated into the service. I bought one for my prusa one, but on the mk4 I use an ESP32 camera board with their firmware. I actually contributed the firmware build for the Freenove rover and designed this case to mount. It doesn't do video, but it was super cheap.
(@ShroudedScribe)
Yeah I've considered octoprint but also thought maybe i just want video only and not to go as deep as having control. Not sure yet. Might start with one and then go deeper. I have a Ender 3 V3 SE and most prints, even long ones, I do, I'm usually around for most of it, and I also don't feel like doing any proxies or exposing any bit of my internal network to the internet (even if securely so), so a camera would suffice in most cases. I suppose it could be neat to have some control via octoprint though, but that might end up being overkill for my needs.
I've just pushed some small updates to my very hobbyist progressive web app Collider. Originally it was "Bat Signal", then "Kid Collider", now just "Collider".
My real life problem is this: I have a kid who wants to go to the park. They want friends to be there, too. I could spend 30 minutes texting various parents in the neighborhood to see if anyone is free and wants to come over, one at a time, wait for responses, haggling on details like arrival time or exact playground or whatever. Finally get something worked out, but now my kid is tired and ready to go home, etc.
What I really want: As i walk to playground I broadcast some signal to friends that "we are going to this park in 10 minutes" and then just go. If another parent is thinking they want to go to a park, too, they can look at this app and see "hey, someone else is going to this particular park, lets meet them there".
Wow, it’s kinda crazy how similar in premise this is to the project I posted. Great minds or something
Wow, i didn't read yours before posting. I made this at the start of the summer but when i shared with friends I hedged a lot, saying "maybe not totally finished, don't use it if you don't want" etc. So not a lot of uptake, and I regret that because all summer there has been this big group text of like 10 people, and only 3 of them are sending out invites etc. A lot of back and forth.
I kept thinking of the old "Check ins" on facebook. I'm not sure if that is still a thing or not, I haven't had facebook in 10 years.
But yeah, we have a lot of the same ideas. I'll sign up for yours and see what else is in there.
The cold start problem is real. And funnily enough the inspiration for mine was born out of old aol instant messenger away messages, but inverted.
Any reason why a "Cool Parents of Gazookville" group chat wouldn't work?
That's what I do and it seems to work pretty well. No one really chats in it, it's just for co-ordination and the occasional meme.
There is a group text of maybe 10 parents that has been the default for the summer.
Some things that fall short:
In my mind, my app is less “loud”…I’m not interjecting with my days plans to a hostage audience. I’m putting a pin on a bulletin board that others can see if they want.
Yea all of that is totally fair. I guess I’m lucky my kids’ play date groups are pretty homogenous. I should have realized that is probably not very common. Good luck with the project!
This is hardly technical, but I noticed tonight that my Magic Trackpad 2's battery puffed up. Seems like a waste to replace the whole thing when I have it plugged in full time.
The little board is for haptics. About five minutes of work and I've got a battery-free magic trackpad 2 that will easily last me another ten years.
I made my daughter (12) solder ESP32 and some electronics components so that she could see for herself that it does something.
The something, that she was making this for, is DIY fans under the radiator to help run air through it when the heat pump is in cooling mode (through the summer) as I didn't want to spend money on proper fancoil opting to leverage my DIY-ness in myself. Thanks to using ESP32 the fans are connected to Home Assistant and I can automate the whole thing AND I can drive them manually, which is what I showed my daughter after she soldered it together and we mounted and tried it.
She was "surprised/amazed" to see it work (as in "My dad wants some reaction"), so I would say she was unmoved by it. Well, I'll try again with some other thing.
Last Thursday I had a sudden spark of inspiration while at work, went home to realize I don’t have it in me to make an app from scratch, and decided to try out bolt.new’s AI tools to spin it up.
I’ve spent the better part of the last week working through issues but think I’ve got most of the kinks worked out. Admittedly, this is still largely a black box that I’m unsure how it functions exactly, but it’s only logging an email so I feel it’s low stakes.
So now for the idea: what if you could post a status that shows you’re available to socialize so that friends, families, groups, social circles in general could see that status and have less friction to reach out? Or maybe even post a block of time that you’ll be available in the next couple of weeks, so that others that may be available at the same time cools reach out and coordinate a group hang.
That’s the basic premise behind Sitch
Add friends, quickly post a status that automatically expires after a set time, and allow the connections to flow more naturally/passively.
Admittedly, it may never take off because who needs another social app? I truly just wish this functionality existed so I could utilize it myself.
I am also open to hearing a critique of the methods used to make this project, I’m certainly not pro AI generally, but it was interesting/cool to be able to take an idea and manifest something so quickly.
I greatly appreciate anyone willing to try it out, provide feedback, even share it with your circle of you find it valuable
Thank you
Collider guy from elsewhere in the thread here.
One thing that I tried to think about a lot was any potential for "harassment", in the sense that I want to avoid some of the perils of networking: cliques, ostracism, etc. The feeling that you are being left out. I'm skeptical that I thought of all the ways people can be mean to each other, but I'm hopeful that my app (and pretty much your app since it's so similar) is so simple in features that I don't have to worry about it. Is that something you were thinking about, too?
Do your friend requests have to be approved by the receiver?
One thing I wanted to avoid was handling any messaging directly in the app. I see you have it in yours. That felt like something I could leave to normal channels of communication, otherwise I'd have to worry about user to user interaction and the social problems that can come up, as well as additional technical debt. Did you consider leaving it out but decide that it was important enough to keep, or did it just feel natural to have it?
I like that your app has future scheduling. I considered it, but ultimately I wanted to keep it as simple as possible. Generally if I'm future planning I am not as bothered about reaching out directly to people to ask if they want to come. If I am scheduling 3 days in advance to go to the zoo, I don't necessarily want to say "anyone is invited" and then get 3 people inviting themselves along, bringing all of their concerns about timing and logistics into it. Future plans are usually more targeted, pre-meditated. Immediate plans, for right now or a few hours in advance, calls for more of a 'hail mary' announcement. There is no time for targeted planning, we are walking out the door now, show up if you want.
Do you see it differently, or given it a lot of thought? I only know how I operate, so it's great to see someone else's perspective.
EDIT: Also, you should try installing your app to your phone home screen. I just did it on my iphone, and the top navigation gets hidden behind the "notch". If you go down this PWA route, you can also add a graphic icon into the PWA manifest so it gets a nice icon when installed.
On harassment, I did give it some thought and reached a position of the friends system along with the temporary nature of the messages would hopefully mitigate the issue. Friends do have to accept a friend request.
The inclusion of messages was born out of my friction with group chats, and hoping to streamline the communication related to meeting up. Again, all the public messages are temporary, and the server gets cleaned up weekly. That isn’t to say that having messages is THE way, but it did feel somewhat natural to include.
As for future planning, I think you can of course still make arrangements directly with people, but sometimes I don’t particularly care who my company is going to be, I’m just trying to cast a wide net to get some socialization in my life.
For example, I have a standing weekly board game night that’s always needing more players. So I put a block up saying the date and time, we’re playing board games, come join.
I do think the simultaneous invention of both Collider and Sitch point to a need, but they do almost fill slightly different niches.
Continuing on my convoluted journey of changing the layout of my smart TV remote. I went to my local hackrrspace and found the correct pads on the back that correspond to the buttons I want to change. I now need to work on soldering wires to these pads so I can then connect buttons to them. Unfortunately I don't have time for that this week as I'm traveling. In the meantime, I got some measurements for the existing buttons and case and now I just need to learn some basic 3D modeling to get a remote shell ready.
Really cool project, and glad to hear you were able to figure out the basics of it. I would suggest before soldering on wires, to plan out the enclosure and the necessary buttons. Do the soldering as the final assembling process. Then, you know how long of wires you need and can also batch all of the soldering in one go. On smaller jobs I find that I am waiting longer for the iron to heat up than I spend actually soldering, so I try and batch my soldering jobs.
Good point, since I've got a lot of time before I can even head back to the hackerspace I'll definitely figure out the enclosure first.
Working on my farming game off and on. I'm now on week 3.
I've worked on implementing buildings: they have a model (that fades if you go behind them), a foundation, a hitbox, and a number of interactable points (for e.g. doors, noticeboards, etc.). Currently this is setup as a Godot scene for each building, but I'd like to later on move it to the data file loading system eventually. I'll do that when I have more than one building though :) I am quite happy with the pixel art house I've made for the T1 farmhouse.
Previously a lot of this interactable stuff was hardcoded ("Press Y to water", "Press E to harvest", "Press Space to open NPC shop interface") style, and it was all tied to the Player class, but this was going to quickly get messy when I had more than 4 actions, and actions that depend on context like building zones. It also was a horribly unintuitive UI but it worked for testing the functionality. Obviously wasn't intended to be permanent but needed to get the foundations in place for context dependent actions.
I'm thinking of dividing actions into three categories: Primary actions (e.g. "talk to an NPC", "open a chest", "read a sign"), Secondary actions ("Open NPC shop UI", "Chest Settings"), and Item Actions ("use the tool you're currently holding", "plant the seed you're currently holding"). I've set up a proper hotbar system to enable this. The other ways games handle it are either like Stardew shops where they split the interactable for the NPC and their other roles (e.g. the "NPC" shop is actually the shop counter), or where the "primary" interaction just opens a menu if there's multiple options. Maybe overthinking what I want the code interface to be for calling that. I don't want it reaching into a bunch of globals for easier implementation of future multiplayer and save/load.
The other next step is save/restore of zone state (e.g. crops placement) which is needed for save/load obviously, but also for zone transitions (like entering buildings). I may throw a main menu and load/save interface in while touching on this anyway.
Finally got copyparty set upon my Mac mini to use as a file server for my local network, and got a custom skin/theme working with it (Catppuccin).
It apparently has OPDS support so I was hoping I could also use it to serve books to my kobo via KoReader (instead of also running the Calibre OPDS content server), but I was getting errors when trying to download to my kobo (of course, the error message didn’t render properly on the Kobo so I’m not exactly sure what the issue was). Might spend some time debugging, or just get Calibre also running on the Mac mini.
I've been setting up a Raspberry Pi 5 as a "home lab". The way the software industry has been headed for awhile now I've been looking at switching to certain self-hosted services (for things where being "in the cloud" literally doesn't matter, especially). Been writing a pyinfra script to set up the machine and install various services. I've been using Claude to write the Python stuff and create systemd unit files, it's quite good at those things and it saves me a lot of time.
My text game journey continues. For those keeping track, I started learning ZIL, the Zork Implementation Language so I could build my own text game based off of one of my novels. Not to be outdone, I decided that just making a text game was not enough, I need a fun way to display it at my next author event. So it was time to make a cyberdeck.
One Nanuk 910 case later and a LOT of 3d printing failures, I now have a complete screen and keyboard deck in a bright red case. The whole thing runs off a FabGL board (LilyTTGO). It's basically an ESP32 that has VGA and PS2 (keyboard + mouse) with an SDcard running CP/M.
Yes. CP/M.
It's... old.
But the machine boots instantly and I have this great CP/M program for running Z-Code text games called Vezza (shout out to Shawn Sijnstra!) that runs all my games and the old Infocom standbys like Zork and Enchanter, etc.
I'm ready for my next author show with something decidedly different. Will it bring attention to the table? Who knows? Did I have fun making it? Yes.
I've also installed Word Star on it so worst case I can swivel it around and write.
Just Nerdy Author Things. :)
There's a blast from the past! In the fifth grade, I wrote a "novel" in Word Star, printed on a daisy wheel printer, and hand illustrated. It was called The Great Fungus Escape. Doubtless you have read it, or seen one of its many adaptations for film, television, and theater, so I won't bore you with the details.
It’s a classic. I believe my first reading was when it arrived at the library of congress. Ah the parades that day were something to remember. All the balloons, the bands, truly magical. Has the fame been arduous?
I can go a few days without getting mobbed on the street, so I'd say it's faded to an acceptable level.
Splendid! I know how it is. Such is the price of fame.
I'm currently getting up to speed on modern web security from the server side and all the proper headers to set as it's been a while since I've had to explicitly think about such things. If anyone has a convenient overview or other resources for what current best practices are I'd welcome it, sometimes it's hard to tell when something's outdated.
I decided to completely yield to AI-assisted coding and see what I could learn. I don't trust cloud providers at all, so I rigged up a small local inference runtime and kilo code, and set to designing the game that I've wanted to make since I was a little kid.
I find the process exhausting; I find I have to be "on" constantly to redirect the tooling back to my intent, especially when there's continuations and concurrency in play. I have to mash the "stop" button way more often than I'd like.
I'll grant that the model I'm using is better at geometry than I am, but it's terrible at creating useful abstractions.
I do note that my test coverage is way better than my personal projects usually achieve.
When I'm done, I'll release the game as OSS.
I've been using the frontier models (Claude 5/ ChatGPT5.6) and found them very capable, but this last week I have started to see the frayed edges of it. I had a much bigger, much more abstract coding task, and I've had to break it down into much smaller pieces to keep the models on task. I find that one feature/vertical is the most effective scope. So I'm not obsolete yet...
I'm having similar experiences. I'm less convinced that it's the model drifting than it is the iterative re-understanding of a problem as the solution gets elaborated.
Even the best horse needs a bit of a tug on the reins occasionally... doubly so as we explore a place we've never been.
I'm still working on porting the PuffinBASIC GWBASIC interpreter from Java to C#.
I've got the interpreter portion functionally done since the code was pretty much 1:1.
Some of the array math functions don't return the exact same results because I'm not sure exactly which algorithms the library they used uses, but it seems those functions were just nice to haves because I haven't seen them mentioned in any of the BASIC resources I've found and they're not used in any of the samples from their repository.
I also still need to clean up and format the code because JavaToCSharp, cool as it is, stripped a whole lot of white space.
Anyway that's the part I really cared about, I want to add some optional scripting portions to a game me and my friend are planning, and JavaScript wouldn't really fit the setting.
The graphics portion however is not exactly 1:1 since it seems to handle keypresses as a queue rather than events.
I could just skip it, but I can't say I'm porting something if I'm only porting half of that thing
Turns out all the code for queuing keypresses was in a class that I commented out so the rest of the project could compile, so still not 1:1, but a lot closer
Maybe not useful to anyone but me, but I've been working on a different type of file explorer. Something like Jupyter meets Haskell, but really aimed at simplicity and safety. Relied on an LLM heavily to get the prototype out. It's useful but clunky in parts. Maybe one neat trick is I can easily ssh tunnel through various servers and play back media. I can even encrypt a remote folder and use it for backups or playing media
https://venus.voyage/
Mostly doing some hygiene and making sure my OPNsense router and Unraid server (and it's containers/etc) are all fully up-to-date. Been a smooth and easy process doing so. Plan on adding some additional stuff to my Syncthing to mirror some media to my server automatically when I download it to my PC.
Also been making changes to my site, moving other sites around because of domain name price increases and such. It's been a good catalyst for me deciding how to reorganize things and what I want to keep and what I want to swap domains on. It seems less common TLDs are increasing in price more over time than their more common counterparts (like com/net/org), so aside from my one .garden address, i'm going to migrate anything else to be a subdomain under it OR to its own com/net/org instead for now.
Continuing to brainstorm ideas for my sites and such. Been doing some pixel art for fun (I'm very much not a visual artist but I like stretching my skillset and brain this way, even just for my own amusement and personal betterment, or simply being a living middle finger to AI every goddamn day of my life going forward).
Been considering diving into learning some programming, which I haven't really done much of in any depth since college many many years ago. I've got some interest in Lua among others, so who knows where that might lead eventually.
Trying out HaikuOS live on a few systems to see how that goes. Same for BSD variants. Just messing with various live forms of distros and seeing if anything interests me so much I want to stick with it on a test machine or my laptop (which is rarely used, so easy to be another test machine).
Might actually buy a few-dollar pi camera to add on to one of my pis for 3D print monitoring.