What have you spent "too much time" trying to fix or streamline?

Going to go with a non-technical answer -- my dryer. My wife mentioned off hand to my dad that our dryer wasn't drying everything we put in it. He suggested that maybe the heating element had gone bad, but I didn't really agree. I figured we were just overloading it or using timed dry vs moisture sensing mode. But my dad loves little projects like that (he's going to be insufferable when he retires!) and offered to help fix it.

And while I love my father dearly, when he decides he's going to help you fix something he's going to pester you about it until it happens. It's all well-intended, and I genuinely appreciate his expertise and willingness to help, but he just wouldn't let it go. So he bought a replacements heating element and we scheduled a Saturday to replace it.

We couldn't find a good youtube video for my specific dryer so we effectively went in blind. He suggests we open it up from the back and start there. But there aren't any screws. I suggest that if there aren't any screws, then maybe you're meant to access it from the front, where there's a conspicuous-looking panel below the door. But he continues down his path.

An hour or two later he'd disconnected all the control panel wiring, pulled the outer casing of the dryer off, and otherwise basically disassembled the whole thing. And wouldn't you know it the heating element is very easily accessible via the front panel I'd pointed out. Replacing the element itself (as well as a few sensors that came with the kit) took about half an hour. Reassembling it took another hour. But to his credit, he rewired and reassembled it correctly first try and the thing worked right away. We were even missing screws!

Now this may seem annoying, but I genuinely learned a lot about the inner workings of dryers and it was a good opportunity to clean out a ton of lint that had built up inside the dryer's casing. So I don't really mind the time "wasted" approaching things the wrong way initially. Plus any time with my dad is worth having because life is busy as hell and we don't get a ton of 1-on-1 time anymore.

But then the dyer stopped working two days later. To be clear, the dryer worked well enough before we "fixed" it. Anyway, my dad suggested checking the sensors he'd replaced. Sure enough, the big heat sensor under the drum failed a continuity test and bypassing it let the dryer start. I replaced the new sensor with the old one and it ran flawlessly for a few weeks.

Then yesterday the dryer stopped heating up. Turns on, spins, runs like it should, just no heat. Surprisingly that does dry clothes ....eventually. We had to take family pictures today, so we ended up drying clothes around a space heater.

Odds are something got shaken loose (there was also a previously unmentioned issue regarding some spade connectors), but I am just done with this dryer. I've been putting off even looking at it because I feel like that's going to turn into a massive time sink and the dryer will die of something else a few days/weeks down the road. If I had $600+ to throw away, I'd just buy another one. Probably a sign to start saving up for one, if nothing else :)

Obsidian. It does so much but...

• There aren't enough keyboard navigation tools for power users
• The task management features are rudimentary
• The metadata you generate is mostly useless without writing more code or using plugins (so more code)

So I:

• write a lot of complex data views, including a weighted connection analysis tool that analyzes documents for a set of tags that meet a particularly regexp and then maps by connection by weight which document is the most important and which important documents connect to it and so on. I used it as rudimentary human network analysis within my employer's business based on the results of interviews I gave.
• many many many table-based databases
• enhancements to the Tasks plugin to support regexp-based filtering of tasks in the plugins domain specific language
  -templates. Just so many templates
• so much custom JS

Ultimately, Obsidian's UX just isn't customizable enough. I found my creation and Obsidian together too overwhelming.

Here, I was hoping to craft a near-perfect tool for knowledge management that could span multiple devices. But Obsidian's performance was generally crap thanks to the plugins. And of course it would be. It's Electron. And the more complex the data structures and the larger the data sets, the slower it gets.

Disappointed.

My hydroponic garden.

I've poured hours and hours of work into research, framing, piping, pumps, power, lighting, fertilizer, etc...

What I have is a series of tubes and buckets filled with mystery water and more basil than I know what to do with.

I've been building an emulator collection that slots into an arcade that I built using an aggregator software called Launchbox and configuring all of the systems and games to work with the control panel I built has been a multi year journey in not achieving anything.

Short answer: pretty much anything I touch.

Some specific examples:

• Recently worked on my cousin's website. And in the design process, I spent hours just tweaking font sizes,
  container placements, testing and tweaking add-ons, etc. There were many instances where I went in to change one small thing, and... Well, changed a lot of other things too.
• In the last couple years I started using "add space after" on Google Docs instead of my previous habit of using double-enter. And also using the indent tool instead of the tab key. I have many drabble docs that still use the old format, and fixing them is a bit of a challenge since Google Docs has limited find/replace capabilities. Also, mobile doesn't let me set either of those things, so I'd have to open the drabbles on my computer to fix the formatting. On that note, I now have two "blank template" docs with the formatting already set
  • Related to above: Back in November I realized when trying to copy some stuff for Timasomo, that while I can find special characters with regular expressions, I can't replace them, and I found out Tildes prefers to have extra spaces. I went as far as to try creating a custom code-based add-on, using some code I found on the internet. It didn't work well though, so I went back and just manually added extra spaces to the snippets I copied.
• Another Google Doc-related struggle: smart quotes and apostrophes on Android are hit or miss. For a long time, they just suddenly didn't work (literally just stopped working one day after automatically converting them for years) and eventually I got used to it, but suddenly they came back. Then I switched to Helioboard, and... Now the struggle has renewed itself. Yay...
• I both love and fear digital art because I will spend hours tweaking small details and cleaning lines. It is a guaranteed time sink for me, hence why I rarely do it. I know with experience I'll get faster, but... Yeah.

Those are just the recent examples off the top of my head, but I've noticed I tend to get pretty caught up in tweaking small details of projects I work on for years now. I figure it's both a strength and a problem, since I can miss some big picture stuff while focusing on the details.

Video files and folders for Plex. Since I am mainly downloading torrents, naming conventions are all over the place, which Plex does a good job of handling but drives me crazy. So assuming it is a show/movie that I plan to keep or watch again, I go in and standardize naming conventions, move all files into the same folder for seasons. My wife downloads a lot but she could also careless so then I am constantly checking with her if stuff she has downloaded can be deleted or if it needs to be fixed. And all of this serves no real benefit, except it makes my lizard brain happy. Plex works fine with 99% of naming formats.

Mine's not a tech problem. I tried to convert my old diesel one ton to run on used veggie oil. Spent WAY too much time and money on it.

It involved putting plastic tanks in the bed of the truck and then running new fuel lines to the injector pump of the engine. Which wouldn't be such a huge deal except that I wanted to have a system where the truck starts on regular diesel fuel til it warms up and then switches over to veggie when its warmed, as you cant inject cold veggie or it'll stall your engine.

Which means everything has to be insulated and heated, including the tank, the lines, and the filter that is positioned just before the injector. And I thought Id be the rocket scientist and do that all with heating elements run off the alternator. By this time Id already spent weeks working on it in my driveway with a lot of trial and error. Experimented and found that an old Cummins 12 valve will run on new veggie oil, transmission fluid, hydraulic fluid, diluted engine oil AND used veggie oil.

I actually got it working enough that I also decided to put extra tanks in the bed of the truck and then added a filtering system and a suction system - that allowed me to pull up behind a local convenience store where I had permission to suck the used veggie oil right out of their disposal tank, through my heavy duty sump pump and into my first tank. From there I filtered it into a second tank. And from there I mixed it with a bit of diesel to dilute it a bit (in winter) to a third tank.

The whole crazy project took me about 6 weeks of daily work to get it running half decently. But here's what no one who does this really mentions: Used veggie oil is a royal pain in the ass to handle. It not only drips on things, its sticky as hell. So pretty soon ALL my tools were covered and sticky, the truck bed was sticky, my driveway was sticky and the pump and lines were kept in a box that inevitably dripped and got sticky too. Its miserable stuff.

It did work. For awhile. The servos to switch over the lines from diesel to veggie were expensive, and the lines and pumps and other components probably cost about $2000 all tolled even though I scavenged a lot of components. In the end I ran it for a few months and then in the dead of a cold Canadian winter the heaters couldn't keep up and the lines gummed up and the engine stalled, leaving me with one very frustrated wife on the side of a barren highway. Cost savings vs expenses I think I ended up at about negative $1500.

I gave up, sold the truck for basically what I paid for it with all the veggie stuff still attached. I dont think the kid who bought cared that it could run on veggie oil and I no longer cared either.

I have a certain degree of ocd so I try to streamline everything. All from making a sandwich in a certain way that I have found out is the most efficient to the way I get out of bed and get dressed to which thing I do first when I shower so that the floor gets the least wet etc etc.. almost everything also has a very specific placement in my apartment or I can’t concentrate on anything else. So yeah all this compounded and much much more has taken way too much time

Oef ... a lot of technical programming projects in the planning phase. I have been walking around with an idea of using a STT (speech to text) engine and LLM to create somewhat of a voice dication/idle thought recorder that on input will untangle a bit of my ramblings before storing away a note.

Now, using openAI apis I could easily put something together that works. But, I don't want to use paid providers so I need to self-host or run it locally. Which is possible because models like openAIs Whisper are open source and available.
So down the rabbit hole we go. Do we use whisper directly? That involves setting up a python environment, installing everything and then can just use it from the command line. Oh, it turns out there is also faster-whisper it sounds like I might want to use that. As it turns out there is even a nice server API implementation someone made, great! We can get started, right? Well yeah... but I'd like things to be timestamped and whisper can't handle it very well if multiple people talk. I wonder how others are dealing with that, ah whisperX is a thing, providing word-level timestamps and speaker diarization, nice! But now I need to figure out the API situation again, alright let's dive into that.

Meanwhile, I probably could have gotten a perfectly fine PoC going if I had just set up a python environment, installed Whisper or faster-Whisper and written a simple endpoint wrapping around the command line interface of either.

Where should I even start :-D

For example: Since around 2013 I have made thermometer to show temps insie the house and outside. Since then I spent 10 years wotking on enhancing this. I went through 7-segment LED display to LCD display, from Arduino as a driver to Raspberry Pi, from simple C programming to complex Python one (PyGame on RPi showitng stuff on the LCD), made my own UI, prepared to add some functions that never made it to exist, tried many ways to get outside temperature wirelessly - 433MHz transmitter (worked really poorly), ESP8266 as transmitter (worked well but needed solar power to charge battery and charging circuit eventually failed to the whole thing got zapped by 20V from panel), tried many humidity sensor none of which worked in freezing temperatures (showed 100% all the time). The literal last drop was when I started Home Assistant and realized there is no place for such custom made thing when you have this kind of service that takes care of basically everythung for you. I bought Zigbee thermometers, hooked some of my own temperature aensors into it and left the idea of my own UI. I finally streamlined it by letting everything go and using already existing solution. Which I should have done long time ago...

In 2020 a used car dealer ripped me off big time on a 2003 Tacoma. In an effort to keep the truck on the road and because I'm fucking stubborn, I've put nearly as much money into it as I originally paid for it, as well as a LOT of time including:

• Buying a welder, learning to weld, and fixing the heavily rusted rear half of the frame, largely with Safe-T-Cap products from autorust.com. Bought a new replacement crossmember for the fuel tank and new hitch mount. All rusty surfaces deflaked and POR-15'd per their arduous three step process. Rust encapsulator products poured and sprayed inside the frame rails.
• All new rear suspension - leafs, hangers, shocks, u bolts, plates, bump stops.
• Rebuilt rear axles (paid a shop to do this, I don't have a press or space for one or the specific-purpose tool) to get new bearings and backing plates installed.
• Every brake component rear of the cab, including learning to flare lines and making all new ones. On the front, new calipers, rotors, pads, some new lines.
• Every single component of the fuel system rear of the engine.
• Every light bulb and exterior light housing except the third brake light replaced.
• Tons of misc: front mud flaps, new radio, fumoto valve, diff and tcase oil changed, new plugs, new spark plugs, new fasteners for almost every one I took out.

In progress:

• I've been trying to make the 4wd work again, but after replacing the tcase actuator and front diff actuator still nothing. Pretty sure it's an issue with some sensor or wiring harness or the 4wd computer. Took the truck to a Toyota dealer and another independent shop but neither could tell me what the problem was, both wanted to drop the tcase which would cost a lot and imo probably lead nowhere.
• The second shop pointed out that the crossmember that the front LCA attaches to is nearly rusted out, but it appears that Toyota doesn't make that part any more, and I didn't have any luck finding it online. Most of these in junkyards will be quite rusty too.

Not in progress, but going to need very soon:

• Cab mounts are very bad, need to cut them off and weld new ones on. Got new bushings, need to find new bolts but no luck so far.
• Exhaust is on its last legs, need to order and install new. Should be easy and not too expensive.
• To get rear mud flaps installed I need to order these expensive brackets from Toyota.
• The bed floor is rusting out, pretty bad now in a few spots.
• The original wheels are starting to look pretty bad, need at least to refinish them somehow. The tires have a ton of tread but are like 10 years old.
• There are big scratches on the doors and roof, need to see if this touch-up paint can fix them.
• The front suspension needs to be done, probably with some lift components to match the rear.
• The radio stopped working for like a year but that's magically healed.
• I no longer like the aftermarket light fixtures and grille and badges on the outside, want to go back to stock look.

So yeah I think I'm selling this guy. It's fun as a project, and I learned so much from owning it, but what I really need is a reliable truck so I can do my other projects.

My life -_-

However, since this will last as long as I live, it is definitely worth it to improve the quality of the time I spend before I return to the dirt.

Right this second I'm in the middle of deliberating with myself on whether I should just use the HandBrake settings I've landed on for transcoding my DVD rip of The World at War so I can throw it on to my Jellyfin server. I've been fiddling with settings for almost a week, even finding a command line tool (https://github.com/alexheretic/ab-av1 with its crf-search command) that lets you automate the fiddling of settings.

My latest test run gave me a file around 45% the size of the raw rip, but I'm used to better size savings. However, The World at War's footage is very grainy film, which is a lot hard harder to compress, so I'm not sure how much better I could reasonably do. Every full transcode of an hour-long episode takes multiple hours to complete, so if I wish to continue fiddling, I'll be waiting a while. At this point, I think I'll just use what I have.

The most recent bug at work.

In a nutshell, I have an excel file. In column A, each row corresponds to a line of XML code. I created a VBA macro to iterate through each line, update values, and write it to an XML file. Then I use a tool in our application to upload the XML and create a bunch of assignments in our software.

The XML it created didn't have any errors parsing, but for some reason the assignments it created had some corrupted artifacts. After a lot of trial and error, I found something very weird.

If I instead copied everything from column A onto column B, and used the XML from column B, everything suddenly worked fine. If I literally copy-pasted what was then on column B back onto column A, and used the XML from column A - nope, didn't work again.

I tested this multiple times. Same results. I copied into two XML files, one from column A, and one from column B, and compared line by line. Same exact code. I even put it through AI to see if there were ANY differences, including things like leading or trailing spaces or anything like that. Zero difference between the files. Still, the column A file works and the column B file didn't.

After a full day of examining this, I simply updated the macro to copy everything on column A onto column B and use that.

Trying to get Steam's in-home streaming to work for my specific needs. I took the effort to run a 100 foot Ethernet cable to my office where my gaming PC lives and hardwired my Steam Deck dock in the living room which is connected to my nice TV. Basic in home streaming works fine. I can play games sitting on my couch with controller or mouse and keyboard, but what I really want is full compatibility with all my peripherals. I have a Logitech G923 racing wheel and the Steam Deck won't recognize it. Next is the Valve Index which in my mind, I thought this should theoretically work. People have been playing Meta Quest games wirelessly for quite some time but apparently this just won't work with in home streaming and Valves hardware. The deck also won't recognize the VR hardware at all unless I do some more customizing that just isn't worth it at this point.

Now, I could invest in some really long displayport and USB cables. I priced this at a little over $100 but I don't want to run more cables, drilling more holes, just for VR or a wheel. I will continue to just lug the PC into the living room.

I tried to reencode my entire audiobook library to opus. All of my audiobooks are in an m4b container, and almost all use AAC encoding. Most are around 64 kbps stereo. AAC is a very good format, but it's a bit old now, and there are some better formats. Opus is a codec designed for spoken audio, but still good enough for decently high quality audio. Basically it is perfectly designed for audiobooks. And it does this all at much lower bitrates than AAC or MP3. I did some ABX testing for myself and settled on 24 or 32 kbps (can't remember which). That was actually a higher bitrate than I needed, but I wanted to err on the side of higher bitrate. Even with this higher than needed bitrate, it gave some insane file size reduction. A 1GB audiobook could get down to around 350MB, without any noticeable change in audio quality. And I finally coerced ffmpeg to actually put opus audio into an m4b container. It seems it would default an m4b file to an mp4 container specific to old iPods. I don't care about compatibility with iPods, so I had to force it to use a generic mp4 container that wasn't restricted to iPods. Anyway I spent a ton of time working on this and successfully compacted my 280 GB audiobook library to less than 100 GB. And then I went to load them into Overcast, which I use for my bedtime books. And it is supremely broken with my opus codec in m4b container files. It would play with some issues, but when skipping it would skip randomly and only ever move forwards. There was literally no way to go backwards. So I had to restore my backup to all my huge AAC book files. All that time wasted. Oh well.

Some fun facts about my library. For a long time my largest book was Atlas Shrugged (never actually read it though) at 63 hours and 1.8 GB. I have the Phil Dragash LOTR books transcoded from mp3 to AAC at way too high of a bitrate, and have never bothered to reencode them. They clock in at around 15 hours and 2.5 GB each. They could be reencoded to probably be about 600-800 MB each without quality loss, so I don't count them. Now my biggest book is Stephen Fry's reading of the complete Sherlock Holmes collection. That one clocks in at 72 hours and 4.11 GB!

I've been working on my HVAC system for a couple years now, and I have finally landed on something that works well for my house. We had an oil furnace on 3 zones and an outside AC unit, the whole system is forced air. Pretty common in the US. Because of limited locations to put many heat pump units outside, and also wanting to try and use our existing ducting, we settled on one 5 ton heat pump for everything and a high efficiency propane furnace for back up. The heat pump works down to -4 F, but I have it shutting off at 25F right now. Then the burner comes on to warm the house.

The math I'm working on now is to figure out the most efficient temperature to switch over to propane. The one thing I'm not trying to put into the equation is heat loss and air tightness of the house. Mainly because I don't have the equipment to figure this out, and it would be expensive. So I'm using efficiency charts for the heat pump/coil/propane burner combo to figure this out. Based on the math it says even down to 5F it says that the heat pump is still more cost effective for me, even with using more KWs.

So at the moment I'm gathering some data from my Ecobee thermostats on running the heat pump at 25F and running the propane at 25F and measuring how long each are on during an hour to see how they are doing at raising the temp in the house. Hopefully I can keep dropping it down and testing. I'm sure people will jump in and say defrost cycles will use a ton of energy below 32F but from what I have read, those are built into the COP from the manufacturer.

Long story short I wish that installers of this stuff would work out the best efficiency for a hybrid setup like this. However, they are more prone to have it change over at 35F to be safe. I spoke to many installers and they all said the same thing. Its probably a safe option for them that prevents service call backs.

My Linux install is an ever evolving construction site. By this point I have spent a lot more time messing with configuration files in a fruitless quest to find the "most effective" way to do things or going back from scratch to try something new than using my computer for actual tasks and this shows no sign of stopping anytime soon.

I used to do this kind of stuff with wayyyy too many things. It resulted in 20 years of half finished projects, and even the projects I did finish resulted in being quickly abandoned for the next interesting version of that project.

I pretty much gave all of that up and it's been way better for my mental health. I use the default version of basically everything, and if I have a question of what app, or tool, or solution to whatever my problem is, I just find out what everyone else is doing and do that. It may not be the cheapest, or most optimal, or most ideal solution, but I've found my time and mental energy is more valuable than would be gained by dedicating the research, planning, building, and learning anymore.

Simpler is almost always better for me.