marklar423

My company moved EiBs of data off of tape a few years ago. It was reliable and durable, but the problem was read speed.

It took so long to move tapes around and read the sequentially (no random access!), and as the data corpus grew it got harder to have a practical backup, even though the data was still theoretically extant.

Do you think it's possible to make GUI software with a Unix philosophy? Specifically piping together small programs seems natural in a shell but I've struggled to figure out how it could work for GUI apps.

I'm with you, I use Pocket all the time on my Kobo as well - I need to cobble together some self-hosted alternative. Did you find another alternative besides Omnivore?

It's funny, I noticed I happening and thought it was proof of the opposite - that there had to be some artificial cause for the supernovae (including the Sun), because a real supernova takes many years to progress, not 20 minutes.

Even after visiting the Sun Station I didn't believe it and thought it was a narrative red herring....so the ending was a surprise to me. Somehow.

Good article, I think this succinctly says what I've struggled to explain to my own relatives:

> The vast majority of people do not know what a Kubernetes is; 99% my skills and knowledge are completely unnecessary to be Good At Computers. Turn it off and on, click suspicious UI button, plug in cable, open Google – these are the only requirements of the job.

However, the bits about Tensorflow and nmap makes me lean a tad towards https://xkcd.com/2501/. I think it's hard to tell exactly where the line between Good and Bad At Computers is.

Adding on to what others said - I added a button that can download book files from my Google Drive, and it was as simple as copying over rclone and a one line bash script.

Kobo is definitely superior to Kindle in my book. The hardware is great and it's completely open to sideloading and even modding (it runs Linux under the hood).

I can't speak to their digital bookstore but they integrate with Overdrive for library borrowing.

What sort of benefits come from getting a Masters? Everyone I talk to seems to say a Masters in CompSci is useless, and that you may as well do a PhD instead.

I worked on a large makeup website for a while, and it was common for us to place test orders and subsequently cancel them. The dev doing it usually put their name and the office address as the shipping address.

One day, we got a big box of makeup at the office - turns out their warehouse had gotten so fast that in the time it took us to cancel the order they had already picked & shipped it.

After that we put "TEST DO NOT SHIP TEST DO NOT SHIP" as the shipping name and "123 Fake St" as the address to try and avert this scenario, which seemed to work until we could modify the system to hide those orders from the warehouse entirely.

I think a good compromise is having metrics that are always on, with the ability to enable more verbose debugging logs as needed.

One of my favorite hard-scifi movies ever.

Anybody know why this has to be a law? Whats stopping each school or the DOE from making their own rule, just like all other school rules?

Like Y2K, there's an argument that diligence from the tech crowd prevented this.

It appears to me there's probably a "best of both worlds" compromise:

1. Send the astronauts home on Dragon

2. Certify Starliner for human flight if it can make it home safely (uncrewed)

If the capsule can make it down, it doesn't matter if there were people in it or not, right?

Part of the problem is sometimes the company doesn't have the license to release (i.e. "redistribute") the server code. In that case they're stuck - the law requires them to both release and not release the code.

Or what about scenarios where the company doesn't have code access to a critical dependency? It's not so unusual either - using a non-OSS DB or cloud service would qualify.

I think a better version of the law should mirror right-to-repair efforts: service providers have to release an API spec and not block attempts to point the client code at the new server, analogous to improving the "repairability" of the software with third party components. Constraining this event to when the service shuts down should mitigate economic concerns for companies.

The full text of the story is on his blog here https://paulocoelhoblog.com/2019/03/30/001/

B&H most famously closes the entire website

https://zeromq.org/ -> TIL really cool, thanks for the pointer.

Mostly because sockets are supported by everything today, and they're easy to understand. What you're describing would certainly work but it looks similar to what the OP did in the blog post, with all the complexity it comes with.

With all this effort required (as the author points out), I start to wonder if a better solution is to communicate via RPC over local sockets.

There will be some overhead, but it might be a wash considering calling over a FFI often involves similar overhead to marshall / unmarshall objects. And the simplicity gains would be massive.

This is a real problem that needs solving, but I can see this sort of thing at scale leading big tech companies to not allow their employees to do this anymore.

Several of the examples reference "deep linking", for example "While we welcome all normal links to our sites we regard 'deep linking' as an infringement of our copyright." - I wonder what that could mean. Anything besides the home page?

Just curious, when you say you made it, you mean from scratch with flour & water? What's your recipe?

I'm skeptical that AI will automate away any truly valuable labor, but even assuming it will - what do you think is different this time compared to say, the Industrial Revolution?

That period rapid saw automation and reduction of cost of many, many peoples' livelihoods, but society is unarguably better for it. Eventually people were freed up to do higher-order work.

It's basically just writing a design document; Google has a similar culture.

I suppose if you're not actually showing it to anybody it becomes rubber ducking via design document.

There's a cross-platform option that will give your users the best experience but will also be the most amount of work: make a separate - native - UI per platform.

By this I mean put as much of your core logic and functionality as possible into a shared library, and then make a separate Swift/WinForms/whatever codebase for the UI of each native platform you want to target. Each UI app would call into your shared library.

There is a bit of duplicate work in making separate UIs, but the UX is far superior (IMO) and you have the advantage of not needing workarounds or plugins to access platform-specific features.

When they started, I understood them as a CDN - storing copies of your static(ish) bytes on lots of servers distributed globally.

Happens to be that a globally distributed network of endpoints is also pretty great for absorbing DDOS attacks, so they started offering that too. And then I guess they parlayed that further into offering other T1/T2 level Internet services too.

I don't use them for CDN / DDOS but I do use them for other things and I think they're great.

It's a semantic thing and I agree with the feedback - specifically with the "F" in FOSS. That sort of license would be Open Source but not completely Free (as in freedom, not beer).

Over the years I've come to agree with this POV, and distilled it down to this:

If your goal is profit, don't open source your core product.

We've seen this time and time again where a company releases their core software under a permissive license, and then bigger competitors come along and resell a solution redistributing their software.

If the company's central goal is profit, this is an existential threat. However if your company goal is to ensure the software exists (as a non-profit steward), this is a resounding success!

This doesn't apply to software that's secondary to your core product, e.g. a useful tool you've developed but don't make money directly selling to others.

What I think the GP means is previously, GFS ran on their own machines and didn't need to participate in the Google-wide quota system for CPU/RAM.

However Colossus _does_ run on the general Google compute infrastructure, and so needed a way to get CPU/RAM quota where none existed (because it was all used by existing, non-storage users).