1,554 karma · joined February 12, 2015
Longshot: go all in on bringing mainline Linux to the desktop/mobile. Linux systems are the only place where Mozilla is the default. Hyper focus on high performance with Linux systems. Memory footprint and battery drain are abysmal. The way security controls for extensions are implemented requires stupid amounts of memory.
Just to be clear I'm not trying to shit on your product. In fact, if your platform allowed me to participate with real money I would 100% use it.
What do I get for buying access to the service?
The idea is cool, but I’m never gonna put money into something where there’s ZERO documentation for what happens to my money once it’s in your pocket.
Yeah there’s the downside. The interface is a bit unwieldy. Documentation for lxd is far lower quality compared to the massive and mature docker ecosystem.
However if you know a lot about what docker does at a lower level, that knowledge is transferable to something like lxd. Most of these containerization tools are all using the same Linux kernel features under the hood!
LXD container instances are much better suited for portable dev environments. They are persistent by default and are intended to run multiple processes within a container instance. They can also be copied/migrated between lxd agents.
In contrast, docker was created to run a single ephemeral process per container. It's nice in production environments, but its also a huge pain for some tasks. That's why were seeing folks create things like docker s6-overlay.
I am hesitant to commit to a tunnel-based approach because where I live I get frequent power/internet outages. I feel that tunneling is something I would explore if my application grows to the point where I would need to rent space in a colocation.
The worst part of shotgun parsing and loosely defined input structure is the difficulty of remediation. I constantly receive pushback from dev teams when I ask them to use regex-based validation per field. What sounds like a simple task actually becomes extremely difficult because lots of apps populate datasets via convoluted monolithic endpoints. Dev teams would have to change the way in which shared services structure and output information. Those shared services are frequently maintained by other teams and any other application which consume the same data would also need to be modified.
In the end, it becomes a compromise where the ad-hoc parsing is tightened/modified to be "good enough". This bubblegum/duct-tape fix only further cements the ad-hoc parsing throughout the org.
Further, some of those 23% are bound to be temporary situations. My initial gut reaction to your link just says "whenever this census was conducted 23% of minors had only one adult in the house."
Not trying to downplay the issue of single motherhood, but the statistics available just seem fuzzy.
The study he sites at https://www.statista.com/statistics/276025/us-percentage-of-... is titled "Percentage of births to unmarried women in the United States from 1980 to 2019"
I think it's quite dishonest to present this data as "2/5 men are raised without a father." I simply do not believe that claim. Maybe my perception of the country is detached from reality so please correct me if my intuition is wrong.
How do you even quantify the intensity of a normal human emotion? It’s all relative to an individuals experience.
Human beings are fucked in the head. I believe it’s always been that way and always will be.
A $30 linode box has like 2 vcpus and maybe 4Gb ram.
Where I live I can get a 1U slot in a shared colo rack for $30-$60/mo. Buy a used dual Xeon blade for a few hundred bucks and now I have a setup with 20x the resources. But yeah I admit there’s a lot more manual effort involved.
Take the recent log4j vulnerabilities. Serving static content and logging trivial fields like request headers would lead to RCE. If that box can route to my home router, and my router has a defect available through routersploit, my network is completely pwned.
A network isolated VM with a tunnel to a remote vps would stop that particular attack.
All that being said…if a sophisticated adversary is targeting me I have to concede there are much easier routes to take.
I’m a security engineer at my day job so I may have conditioned myself into excessive fear.
I would get worried about somehow enabling access to defects in my router by opening some inbound ports. I realize that's a little paranoid...but recently I have been playing around with https://github.com/threat9/routersploit and routinely find defects in consumer routers.
Here's my other beef with cloudflare: Once I gotta pay 200+/mo for their security services or whatever, I could just rent out a private rack in a colocation and throw some old beefy lga-2011 xeon hosts. Now I don't need anything on my LAN exposed and I have dedicated IPs, physical security, and backup generators...etc.
An easy secure setup would be to spin up a guest VM and isolate it in its own subnet.
Disable routing between your guest and the rest of your lan and you can sleep easy at night so long as your app doesn’t serve any crazy dynamic content.
Dependency management:
Go mods is a dumpster fire. `go get` and `go install` is finicky and inconsistent across systems.
It's difficult to import local code as a dependency. Using mods with replace feels like a shitty hack, and requires me to maintain a public repo for something I may not want to be public. I end up using ANOTHER hack that replaces mod references to private repos and I have to mess with my git config to properly authenticate.
I've never used another language that made it so difficult to import local code. Rust's cargo is so much easier to use!
Sane dynamic json parsing:
Having to create a perfectly specified struct for every single json object I need to touch is terrible UX. Using `map[string]interface{}` is just gross.
Again, I think Go should copy the existing rust solution from serde. With serde, I define the struct I need, and when I parse an object, the extra fields just get thrown out.
If anyone thinks I'm misunderstanding something, please enlighten me. I hope reasonable solutions already exist and I just haven't found them yet.
Nevertheless, good luck and I hope they don't shut you down too fast.
XMRs randomx pow algorithm is a step in the right direction for sustainability and decentralization.
Inb4 “pow actually has a larger barrier to entry!”
Consider this: at least mining operations have to cover their op expenses by selling their rewards.
With pos, there’s an incentive to put your rewards back into your validator pool further restricting supply for the general public.
I have a homelab too but getting “enterprise grade” service from comcast seems to be my biggest barrier to scaling without leaning on aws.
Platforms gotta monetize by gathering/selling user data. Developers want abstractions to make programming more pleasant.
It’s kind of hilarious that our advances in HW just means that our software resource usage expands to fill the gap.