Programming thread

[306h4Ge5eJUJ]

I found out the hard way that they literally have no customer service department.

Google's customer service department is writing an angry blogpost about how Google mistreated you and getting it to trend on Hackernews. I wish this was an exaggeration.

 

[Jarry Seinfelt]

Google's customer service department is writing an angry blogpost about how Google mistreated you and getting it to trend on Hackernews. I wish this was an exaggeration.

Your company can be writing monthly checks to Google for millions or tens of millions, and when GCP shits the bed, their "round-the-clock service" consists of a support engineer emailing you that they don't have any updates, every hour, on the hour. This can go on for weeks. It makes a lot of sense, as when you're that locked into their ecosystem they've got you by the nuts and don't really care.

One benefit of working for boomer megacorps is that up until 2018 or so, they all had their own dedicated server farms, so any support types were getting their paychecks from the same place as you. And if they gave you the runaround like cloud providers do, you and your boss just went up the chain until they had a SVP breathing down their neck to fix it.

 

[DavidS877]

Your company can be writing monthly checks to Google for millions or tens of millions, and when GCP shits the bed, their "round-the-clock service" consists of a support engineer emailing you that they don't have any updates, every hour, on the hour. This can go on for weeks. It makes a lot of sense, as when you're that locked into their ecosystem they've got you by the nuts and don't really care.

We're hitting a similar thing with a software vendor. Their product has a REST API and an SDK for the language we use.

BUT, any problems with the SDK they claim "Not our problem, it's Open-Source" And by "Open-Source" it means in a github repo owned by them and 95% of all commits are by their employees.

Fuckers.

 

[Jarry Seinfelt]

Apparently some government agencies have contracts with the big cloud providers guaranteeing 100% uptime. Since 100% uptime is not truly possible, the way this works is that AWS or whoever pays the government back for any downtime. I don't have any insight into how that works in practice, as in, how often it actually happens, how much they're paying the gov back in an average year, whatever. Just something to think about.

 

[y a t s]

Well I can also say that from experience, seriously, fuck google. I tried making an smtp script and emailing myself and google locked me out of my email accounts for a month and I found out the hard way that they literally have no customer service department. What in the fuck?

And R. Kelly wrote "I Believe I Can Fly". He's a piece of shit beyond description, but it's still a good song.

 

[${Sandy}]

So a decent vpn generally costs money but I had an idea.

If you just wanted to mask your ip to your isp and whatever site you are visiting cant you just fucking write a script that spoofs your ip?

Like my computer just tells the isp, " NO STALKER CHILD I AM NOT CONNECTING TO PIRACY WEBSITES. YOUR LIFE IS ALREADY OVER STALKER! MY ONLY INTERNET TRAFFIC IS GOING TO 1.867.530.9."

 

[DavidS877]

So a decent vpn generally costs money but I had an idea.

If you just wanted to mask your ip to your isp and whatever site you are visiting cant you just fucking write a script that spoofs your ip?

Like my computer just tells the isp, " NO STALKER CHILD I AM NOT CONNECTING TO PIRACY WEBSITES. YOUR LIFE IS ALREADY OVER STALKER! MY ONLY INTERNET TRAFFIC IS GOING TO 1.867.530.9."

That's great, as long as you don't want any packets to come back.

 

[Kosher Salt]

That's great, as long as you don't want any packets to come back.

In some DDoS (and pizza-based) attacks, you don't want any packets to come back.

 

[std::string]

Apparently some government agencies have contracts with the big cloud providers guaranteeing 100% uptime. Since 100% uptime is not truly possible, the way this works is that AWS or whoever pays the government back for any downtime. I don't have any insight into how that works in practice, as in, how often it actually happens, how much they're paying the gov back in an average year, whatever. Just something to think about.

If you work for a private company with bloodthirsty attorneys you, too, can get the same thing.

Last hospital I worked at we charged around $40k/minute for downtime.

 

[macrodegenerate]

So a decent vpn generally costs money but I had an idea.

If you just wanted to mask your ip to your isp and whatever site you are visiting cant you just fucking write a script that spoofs your ip?

Like my computer just tells the isp, " NO STALKER CHILD I AM NOT CONNECTING TO PIRACY WEBSITES. YOUR LIFE IS ALREADY OVER STALKER! MY ONLY INTERNET TRAFFIC IS GOING TO 1.867.530.9."

That would work if your packet wasn't wrapped around a packet generated by your router. Then wrapped around a packet generated by your ISP's router so far and so forth. At best everything makes it back to your router which will then promptly discard it.
THERE IS NO 1.867.530.9. ON THIS HOME NETWORK. ENJOY PRISON STALKER CHILD.

Edit:I'm wrong.

 

[DavidS877]

That would work if your packet wasn't wrapped around a packet generated by your router. Then wrapped around a packet generated by your ISP's router so far and so forth. At best everything makes it back to your router which will then promptly discard it.
THERE IS NO 1.867.530.9. ON THIS HOME NETWORK. ENJOY PRISON STALKER CHILD.

(Unless you're using a VPN, then it is encapsulated.)

 

[UERISIMILITUDO]

Since 100% uptime is not truly possible, the way this works is that AWS or whoever pays the government back for any downtime.

It is possible, and used to be common. When was the last time fucking Visa or MasterCard went down? As far as I know, they don't, because the system operates more like a mesh network. They also use IBM mainframes, which also have a reputation for not going down, at all; those things can contact IBM for replacement parts when their diagnostics report coming failure, and parts can be replaced without turning off the machine. The Multics hardware had the same characteristic, there's a story about turning off part of the machine gradually and reassembling the pieces into a completely separate unit, because it was meant to be used as a utility for computation no different than water or telephone.

This may be a bit of a stretch, but Bitcoin hasn't gone down either. Sure, it doesn't do much in comparison to some of these other examples, but it has never "gone down" globally, now has it?

Something some of these systems have in common is their age. Back when computers were new, they had to work at least as well as what they were replacing, but now they're common and don't work worth a fuck.

Always be careful when deeming something to be impossible. There are a lot of retards in computing who insist that things be impossible, but it's really just an excuse for their pathetic nature and inability.

 

[Weapons of Mass Tardation]

Your company can be writing monthly checks to Google for millions or tens of millions, and when GCP shits the bed, their "round-the-clock service" consists of a support engineer emailing you that they don't have any updates, every hour, on the hour. This can go on for weeks. It makes a lot of sense, as when you're that locked into their ecosystem they've got you by the nuts and don't really care.

One benefit of working for boomer megacorps is that up until 2018 or so, they all had their own dedicated server farms, so any support types were getting their paychecks from the same place as you. And if they gave you the runaround like cloud providers do, you and your boss just went up the chain until they had a SVP breathing down their neck to fix it.

Yeah, of all of the major cloud providers, GCP hands down is the biggest load of AIDS by far (the only one worse is fucking IBM Cloud). The only thing that supposedly is better sometimes versus Azure or AWS is apparently security, but every time I have to use GCP I want to fill the computer with birdshot. As much as I hate MS, Azure is actually pretty nice to use, and AWS is what it is. I've been a part of a major migration from on-prem to GCP, and god damn was it ugly and expensive as fuck, and generally everything got way shittier. The PHBs were also shittting themselves because the Google sales rep promised it would "cut costs by at least 2x", when in reality they probably doubled at the end of the day, but only after the company was locked into GCP. I remember getting a lot of strange looks a decade ago in tech circles saying Google are was and was going to continue be one of the next great Satans, especially from jeets and soys that thought I was doing a schizophrenic wrongthink, but I feel pretty vindicated.

I really do miss the days of good on-perm. I remember one of the early tech jobs I took many years ago, the IT and sysadmins were in the same building, and they had absolutely amazing infrastructure. You could get anything you needed running, and they were a fun group to go shoot the shit with, and you had someone basically on hand to help do stuff. The most fun part was getting to help work on the physical hardware, racking and unracking systems, switches, big JBODs and SAN disk arrays. A huge amount of the cloud talk is propaganda and a scam, as a large amount of people buying get taken to the cleaners costwise for their workload vs on-prem or colo. All that bullshit talk about getting rid of all your ops people was a big fat lie, you just now need an equal or increased number of expensive "cloudops" and "site reliability engineer" jannies to clean it up in perpetuity.

So a decent vpn generally costs money but I had an idea.

If you just wanted to mask your ip to your isp and whatever site you are visiting cant you just fucking write a script that spoofs your ip?

Like my computer just tells the isp, " NO STALKER CHILD I AM NOT CONNECTING TO PIRACY WEBSITES. YOUR LIFE IS ALREADY OVER STALKER! MY ONLY INTERNET TRAFFIC IS GOING TO 1.867.530.9."

I guess even basic networking knowledge isn't a prerequiste to turn out webshit these days, is it? You should go read the R. Richard Stevens books end to end and try to understand why what you're asking doesn't make sense.

It is possible, and used to be common. When was the last time fucking Visa or MasterCard went down? As far as I know, they don't, because the system operates more like a mesh network. They also use IBM mainframes, which also have a reputation for not going down, at all; those things can contact IBM for replacement parts when their diagnostics report coming failure, and parts can be replaced without turning off the machine. The Multics hardware had the same characteristic, there's a story about turning off part of the machine gradually and reassembling the pieces into a completely separate unit, because it was meant to be used as a utility for computation no different than water or telephone.

This may be a bit of a stretch, but Bitcoin hasn't gone down either. Sure, it doesn't do much in comparison to some of these other examples, but it has never "gone down" globally, now has it?

Something some of these systems have in common is their age. Back when computers were new, they had to work at least as well as what they were replacing, but now they're common and don't work worth a fuck.

Always be careful when deeming something to be impossible. There are a lot of retards in computing who insist that things be impossible, but it's really just an excuse for their pathetic nature and inability.

I remember working with an old timer that got his start on system 360 and 370 at IBM, and learning about all of the stuff those could do. You could replace basically everything on those systems in an online fashion, and yeah ~100% uptime was not impossible. Visa and Mastercard probably still use the descendent systems (zSeries), which also probably gets real close to perfect uptime. I always wanted to build a computer that you could do something like hotswap processors out. But yeah shouldn't ever say impossible; people swing shit like the CAP theorem around without really understanding what it means to say something or other obviously can't be done.

 

[${Sandy}]

I guess even basic networking knowledge isn't a prerequiste to turn out webshit these days, is it?

If you think I'm bad, I heard there are people with computer science degrees that don't even know what an API is.

Also, I thought more about my idea and it sounds dumb, the only way I can think of to mask your ip is setting up your own server to connect to, but at that point you might as well save yourself the effort and just keep paying for a vpn.

You should go read the R. Richard Stevens books end to end and try to understand why what you're asking doesn't make sense.

Well, I want to get better at this shit so Thank you for the suggestion.

 

[Weapons of Mass Tardation]

If you think I'm bad, I heard there are people with computer science degrees that don't even know what an API is.

Also, I thought more about my idea and it sounds dumb, the only way I can think of to mask your ip is setting up your own server to connect to, but at that point you might as well save yourself the effort and just keep paying for a vpn.

Well, I want to get better at this shit so Thank you for the suggestion.

We all started somewhere, you're going the right direction. All of the R. Richard Stevens books, including and especially the TCP/IP books are dense, but really really excellent books. If you want to learn shit about network systems, Computer Networks: A Systems Approach is pretty decent (read through it recently to help a younger developer I'm mentoring), and Beej's networking guides are also good.

 

[macrodegenerate]

It is possible, and used to be common. When was the last time fucking Visa or MasterCard went down? As far as I know, they don't, because the system operates more like a mesh network. They also use IBM mainframes, which also have a reputation for not going down, at all; those things can contact IBM for replacement parts when their diagnostics report coming failure, and parts can be replaced without turning off the machine. The Multics hardware had the same characteristic, there's a story about turning off part of the machine gradually and reassembling the pieces into a completely separate unit, because it was meant to be used as a utility for computation no different than water or telephone.

This may be a bit of a stretch, but Bitcoin hasn't gone down either. Sure, it doesn't do much in comparison to some of these other examples, but it has never "gone down" globally, now has it?

Something some of these systems have in common is their age. Back when computers were new, they had to work at least as well as what they were replacing, but now they're common and don't work worth a fuck.

Always be careful when deeming something to be impossible. There are a lot of retards in computing who insist that things be impossible, but it's really just an excuse for their pathetic nature and inability.

You are not kidding about mainframes. Absolutely fantastic machines. They have entire drawers of CPUs and RAM with redundencies if each of them fail. Most people will never use all of the hardware on a mainframe because of their expensive SLAs, but when unleashed they are absolute units in every way. Easily wiping racks worth of compute with DBMS operations at a fraction of the speed. The latest ones even have hypervisors for Linux VMs built in.

Only problem is the lack of personal to manage them, and low adaption in the market.

 

[${Sandy}]

We all started somewhere, you're going the right direction. All of the R. Richard Stevens books, including and especially the TCP/IP books are dense, but really really excellent books. If you want to learn shit about network systems, Computer Networks: A Systems Approach is pretty decent (read through it recently to help a younger developer I'm mentoring), and Beej's networking guides are also good.

I'll check those out. Most of my learning so far has just been front end stuff that I did some freecodecamp classes for until I had a solid grasp on the basics then I just started making my own shit.

I've been considering pick up a few books because they probably have a treasure trove of knowledge that free learning materials on the Internet don't cover in a meaningful way.

 

[y a t s]

the only way I can think of to mask your ip is setting up your own server to connect to, but at that point you might as well save yourself the effort and just keep paying for a vpn.

On top of that, the benefit of using a public paid VPN service is that it has other users who appear to connect from the same server IP. If you run your own wireguard server off some VPS in Europe, you'll most likely be the only guy appearing to connect through it.

This is mainly why the US Navy made Tor public. They don't want all of their exit nodes to trace back to the military, since that defeats the point.

 

[Belisarius Cawl]

I've been considering pick up a few books because they probably have a treasure trove of knowledge that free learning materials on the Internet don't cover in a meaningful way.

Well yeah but especially for technical books the one-finger mouse click discount applies. I could hammer ls into my e-books folder and give you the output but that's over 5,700 files and unsurprisingly a lot of them are things like 40K novels. What you could do though is give me the broad strokes of what you want to get into and I can make recommendations based on my knowledge and titles I have at least vetted if not read.

 

[macrodegenerate]

I'm reading "design for hackers" and it's not bad.