- I have two moonlanders but wound up giving up because I just couldn't adapt to it. And when writing code, finding the symbols like {, =, }, and other common coding ones was just too difficult to retain muscle memory. I tried many layouts, I tried to make my own, and at the end of the day it was just too different. I wish I had a better experience.
Any recommendations?
- I don't like the Tidbyt v2. I hated it from the minute I unboxed it, would not recommend. The pixels are very blurry and the image is not crisp at all. The v1 was pretty good, looks more like a dot oriented board (far less resolution). more discussion: https://discuss.tidbyt.com/t/gen2-blurry-screen/6654
I replaced the tidbyts with a Dakboard, specifically a https://shop.dakboard.com/products/pixeltoptech
- > I guess really the bigger issue here is the dynamic nature of the prefixes. If I could go log into IANA and click a button and get assigned a /48 then log in to my ISPs' sites and attach it to my connection and be on my way... I'd deploy it today, no problem.
Yes, this right here. It should work like a phone number where you can take it from provider to provider. You setup your gateway with your IPv6 prefix and then go.
Problem is all the technical and non-technical stuff to actually make that work.
- Verifying ownership of the prefix (rPKI or IRRs) and which ASNs are allowed to advertise it.
- Limiting allocations to someone? I guess phones solve this with SIMs
- Getting your ISP to route the traffic to your device
- Getting the device to announce it with RAs
- Dealing with asymmetric return routing over a slow "path" since it could be multi-homed
- Dealing with routing aggregation since IPv6 routing tables would explode with all the /56s in there
Maybe bring back Mobile IPv6? https://en.wikipedia.org/wiki/Mobile_IP
- Tesla build quality is atrocious, the interior feels cheap, and their repair process is very long cycle time. I’ve known several people in the “never (Tesla) again” camp and now they’re buying Mercedes EVs, rivians (surprisingly) and Lucids.
I have had a few Mercedes and they know how to maintain service on cars for a long long time.
I’d never buy a Tesla. They are a joke and you’re paying for the name and they used to be able to charge a premium because they were the first to market
- You get recruited by other VPs to be VP. You skip all the normal interviews that would discover your incompetence. You get hired with a plan of “we will figure out where he/she best fit”.
You take over a team or several teams that add up to a hundred or more people in total.
You probably don’t provide value and just added a layer of management and made vertical communication harder and more broken. You’re leeching off the company.
This brokenness is perceived as a leadership gap and more middle managers/VPs are brought in and the problem is exacerbated and the cycle continues.
Hopefully, someone who actually cares about the company they work for notices they have 40 VPs in an org of 500, or 4 layers of VPs and asks “wtf” and actually does something about it.
One of the VPs leaves, then ruins another company the same way.
Bottom line: VPs should be developed from within, or have such an obvious and clear role they are recruited for, and aren’t allowed to hire their friends or previous coworkers.
- systemd is great.
The first thing I do on a Debian machine is replace everything possible with systemd services.
I know a common criticism is that it does _a lot_ but everything that was done before are bespoke scripts (ifupdown), tools that generate configs for other tools (Netplan, resolvconf), or just completely inadequate for current use cases (sysvinit).
While the core of systemd was to replace sysvinit, upstart, and whatever dark ages we used to have (impossible/bespoke service dependency (remember putting dependencies in init script comments!), scripts that used to inherit environment from whatever user you ran the init script as, scripts checking their own PID files (!), and implementing their own logging and log rotation).
Linux is far far better off with systemd, systemd-resolved, systemd-timesyncd, systemd-networkd, systemd-boot (grub is just a monster).
I also love they're pushing forward with things that have been very very difficult to use like Encrypted FS, Measured/Secure boot, UKI images, udev, etc.
I have a lot of respect for the systemd development team pushing forward their vision in the face of criticism (founded and unfounded) and getting this far without fragmenting the ecosystem even more (like ubuntu does).
- I absolutely love Taskwarrior. I love that it'll just tell you what to do next based on weight, priority, due date, size, etc. I love annotating and adding tags/labels etc and how generic things are.
It's really a pain in the ass that I can't get it on multiple devices so I ultimately never am in the right place to add todo items. I know there's a Task champion sync-server that's rewritten as part of Taskwarrior 3.0 but it seems very early in the development and I haven't gotten to to work and have been using Inkdrop instead; would love to go back to Taskwarrior.
Systems must be _structurally architected_ with security in mind.
Security is layered, using a random key with 128-bit space makes guessing UUIDs infeasible. But _also_ you should be doing AuthZ on the records, and also you should be doing rate limiting on API so they can't be brute forced, either.