I was super bummed out when I thought i could upload my purchased ssl certificate over to heroku on my rails app. Lo and behold they dont do that, and the moment you start paying for a 'dynamo' you get that super sweet ssl.
Though screw that, i have my $5 linode box im planning to begin hosting soon enough...just never get around to it.

the cheapest heroku dyno is like $5/month i think.

Dumb back end question: I have a large json array that I am using for a stats/graph page, previously every 24 hours a JS setinterval function would fire making a mongo query and then holding that data in memory and serving it to the page when someone hits it (ajax), but it seems to make better sense to have a script that I can put in a cronjob that would do the query and then write to disk and just serve that? Or does that not make sense? Its about 20000 JS objects and growing.

if you can write to a flat file and your web server (nginx or apache) serves it statically (ie, without hitting node or anything else), i think you want to do that. that said, serving from memory even through node is very fast and if you can afford the RAM, what you're doing is fine. but as you said, it won't scale.

You don't get free SSL with Heroku, though it seems to state that. You are allowed to use SSL when you pay, and with the SSL access point, you can (probably) use whatever cert you want, though some certs are easier to use than others.

OmgGlutten!; the paid dynos start at $7 / month.

Didn't even think about nginx serving it yeah I'll do that, thanks. Its a $10/month digital ocean server that runs mongo and my node app @ ~30% cpu and last time I did a release it threw an out of memory error so had to axe the entire stats page. So yeah now I have something to do, never done a cronjob in my life.

100 players at once on in my game first time this weekend! badass.

Sorry if this has been posted, I've been away and skimmed the thread: https://np.reddit.com/r/cscareerques...n_database_on/

Basically a junior dev destroys production DB on day 1, backups don't work, CTO fires employee. Apparently there was a manual script to set up your local environment. One of the steps is to provide your local credentials, however the PROD (!!!) credentials are listed in the setup document. Employee accidentally copied and pasted those instead of supplying local credentials. After he ran a script that populates some data to test against - and wiped out the prod DB. LOLOLOL

Best comment:

This reminds me of a big lawsuit that I think actually shut down Martin Luther King Jr. hospital in LA. Some nurse turned down a patient's heart monitor, and the patient died when the alarm went unheard. WHY IN THE HELL DOES A HEART MONITOR HAVE A VOLUME KNOB IN THE FIRST PLACE?

that's awesome, man. congrats.

This is amazing. Sucks the JR dev is the scapegoat here.

Kinda have my doubts that is real, it just reads too much like a first day disaster you'd make up if you had to write one. But if it is, LOL.

And you definitely want a way to be able to shut a heart monitor alarm off because an alarm going off is not conducive to clear thinking or clear communication in an emergency. But it shouldn't be a volume control - it should be like a button that shuts the alarm off for some limited time period.

There are two kinds of database devs:

-- has nuked prod
-- hasn't nuked prod yet

Why would people think this is a fake story?

It could easily be real, nothing in it is unbelievable, but there are some things arguing for it being fake, like in a real story the screwup could have happened on day 2, 3, 4, whatever, while in a fake story it's always going to be day 1.

There are also little details that don't make a lot of sense to me, like this:

The chronology seems wrong here. CTO is telling him that legal might have to get involved because of severe data loss, and only after that does he find out that backups aren't restoring. If they don't know yet that they don't have working backups, why is the CTO losing his ****? If they do know, wouldn't the CTO tell him, as part of this rant about data loss?

Similarly this:

Wait, so you didn't explain what happened in your face to face with the CTO? He didn't ask? Surely they would have gone through with him exactly what he did. What he did was very simple, what would be left to explain?

If I was a distraught 22 year old in this situation, I probably wouldn't have the wherewithal to explain much of anything while talking to an angry CTO, so I could definitely see explaining that later on Slack.

Why would a CTO make legal threats, not know about backups, etc?

First, the legal threat may be more about losing money because there is no functioning database. If the company earns $1k / hour, then that is a hard loss of income.

I also wouldn't assume that a Cx-anything is, by definition, totally clued in. I'm thinking US-centric, but your qualifications for becoming a CEO is paying a few hundred dollars to open an LLC. Hiring employee #1 can be labeled the CTO, regardless of anything.

Yeah, the day 1 thing might be a slight thing, but then again, if they are asking him to set up a dev environment alone on day one... that doesn't even make sense in many places, right? Why would a CTO allow that?

The part I'd be skeptical of, if any, is the idea he moved across country for this one..

My kid was in the hospital for a couple of months when born. It's sort of mind boggling how bad monitoring is of patients. The best practices and tools we have for monitoring computers are light years ahead of what we do for patients.

You need to be able to turn down heart monitors because what's acceptable changes based on circumstances and it can be really distracting and bad for the patient if the alarm is constantly going off and there's nothing to do. Ideally you'd use an adjustable threshold (which I think they might actually have) but I suspect tuning that would actually require a doctor and not a nurse. And no doctor has time to constantly check to make sure each patient is set at the right level.

But in general the whole situation is ****ed up. Each system has their own monitoring and alerting. And the severity of the alert (in terms of noise and visual cues) has little relation to the actual severity of the situation. Something that just requires action in the next 30 minutes might alarm louder than something that requires immediate life or death action.

And with patients in serious medical conditions there's almost always going to be something 'wrong' with them. So nurses end up 'learning' what alarms are usually false positives and which aren't. They rarely react with urgency because there's so many false positives. Etc. Etc.

I could rant about this for hours.

I guess there are probably 'lots' of Cxs that aren't clued in, but its almost certainly the minority (at least for their respective areas) for any non-trivial (say > 4-5 employees and relatively profitable/funded) company. The CTOs of these types of companies have generally done a lot more than just pay a few hundred dollars.


Huh? This is basically the first technical task for any developer at any company I've worked at. Sure, they have a person assigned to them for help/questions, but generally the task is "follow these instructions and set up your development environment". And if we set up those instructions/the vm in such a way that a new employee could easily **** up production - that would be completely on us and not them.

Again, why? I moved to a different country for my first job. Lots of people move from the area where they grew up / area they went to school to a new place right after graduation.

The whole thing could certainly be fake. But its also definitely believable.

I can totally see the CTO losing his **** and firing the kid without any more information than the prod database is gone, and I'm not projecting this on former employers at all...

Not saying this is exactly your situation, but I always have to remind myself that the nurses and doctors are seeing patients all day every day, and what is huge to us on an individual scale is mundane work for them on a large scale.

FWIW, MLK hospital was shut down for many violations, but famously, this...

The center attracted national attention in May after a homeless patient who had come through the emergency room collapsed on the floor, screaming in pain, but got no response from employees. A janitor mopped up around the patient as she vomited blood. She later died.

http://www.nytimes.com/2007/08/11/us/11hospital.html

My post was overly dismissive, and that wasn't my intent. In places like SF with their apparent bias to younger people, the CTOs, CEOs, COOs, are often going to be a little more fresh-faced.

It was more a response to ChrisV's skepticism that a CTO wouldn't know any better.

Oddly, I've only seen to opposite.

I certainly see the value in letting devs have the freedom to build up a machine the way they see fit, but at the same time, how do you deal with programming language / database version conflicts, etc?

Here's an example I actually did have to deal with. I used PgAdmin and the team I was working with used SQuirreL. This wouldn't be an "issue" on the surface, until you realize that SQuirreL doesn't support dollar-quoting and other needed features to work with Postgres correctly. We both had things that worked, but we couldn't actually run each other's code using our GUIs, and in fact some of their stuff wouldn't work via the CLI because SQuirreL added whatever atrocious things to work-around it's own bugs.

Also, why wouldn't they, at the very least, use a bash script to automate the on-boarding process?

From the description, using the prod database, giving a fresh-faced admin access, etc, like that, is plain bad practices, and presumably, any CTO should know better...

Because the standard for getting a job as a relo tends to be much higher than the standard for getting a job as a local, but that obviously depends on where you are planning to move to.

Are there AIs in practice, or being developed that will take existing working code, and refactor it into more layers of abstraction?

Seems like eventually you could have humans writing the main logic and then just telling the AI how to make it more modular and do a lot of the "busy work" to make the code better and more robust, add tests to it, etc.

Employee who deleted database was kicked-out of hell, imo. Terrible company.

This isn't my point. My point is that the way we monitor patients is ABSOLUTELY (like not even close) causing worse health outcomes and more deaths than if we monitored them with the current best practices/tools that we use to monitor computers.

There are many reasons for this (some stupid and some legitimately challenging to overcome). But the reality is that I suspect the number of people that actually understand this is tiny. And of course we can rarely point to the exact people impacted by this.

For example, let's say the response time to a heart rate going to 0 is directly correlated with the patient's chance of death or suffering other complications (reasonable assumption). If the many false positives of a heart rate alarm cause the average response time to increase by some non-trivial amount (definitely happens) then we're having people die 'unnecessarily' but there's not really individual patients we can point to because there are still going to be deaths even with the decreased response time.

Edit: And to be clear, my point isn't a criticism of the people in the process. It's the process/tools that are the issue and the people generally do the best they can in that system.

I don't think we're talking the same thing. I'm talking about having a developer set up their development environment which would (ideally) match the test/staging/development/production/whatever-you-want-to-call-it environments.

So I imagine this situation was more similar to "Install this version of a database" (or run this set container/vm/whatever where it is already set up for you) and then step 2 is to populate that database.



I mean, you can always automate more. In this case I suspect there's a general purpose script that tears down and rebuilds your database to a known good state. That's something you'd want to be able to do at least semi-regularly. So the on-boarding process is probably about setting up the environment as well as getting the developer use to the tools/scripts they can use during development.

Yeah. I think this is what everyone agrees on. The setup script shouldn't have had production credentials. A local developer probably shouldn't have easy access to the production machine. And a brand new developer probably shouldn't have any access at all to the production database.

Based on a quick Google search it seems the idea isn't new but only recently with our computing power and more people working on AI a lot of the big players (Microsoft, Google, MIT, etc) have made some limited progress with AI writing code.

Sent from my SM-G900R4 using Tapatalk

Apparently my work VPN stopped banning 2p2 (category: gambling). The office network was weird - it would ban like every 4th hit during the day - increasing to every hit by late afternoon. Must have been sharing the load or something.

I'd like to think I had a hand in it being un-banned. They saw like 500 requests a day come in and decided to take a closer look at the site.

I self-banned from the forum for a while, and still have quit politics. W/o the 2p2 politics forum, I now have an extra 2-6 hours per day to do other stuff. That is not an exaggeration. If anything it might be low. This thread is the main reason I don't quit entirely. It's a great resource.

About one year ago, I went into a hospital with a heart-rate of 180/x and a body temp of 92 degrees. I was unable to sit up without being strapped to the wheel chair. They simply sat me in the waiting room and tossed a blanket over me.

I feel like I had higher priority than people who were throwing up, but on the other hand, I zomblie-lurched out 6 hours later, declining treatment. To me, this was HUGE worry, but to them, it was just another day of whatever.

Similarly, there was a poster in the Lounge talking about his wife's second miscarriage. She was bleeding like stuck pig in the waiting room, but they didn't really pay her much attention. I'm not sure the entire detail, but I think about 8 hours later, they checked her for a fever and sent her home.

I feel like medicine and tech should go hand-in-hand, but I simply don't know what improvements are really needed. As an observer, it looks incredibly broken to me, but what do I really know about that?

I'm saying I never actually seen a developer set up their own machine. Granted, my sample is small, but never saw it. I could see this being more common in a newer company, but I feel like it'd be uncommon in more established companies.

Every job I've ever had the first task for the dev is to set up their own machine.

When I first started at my current job the job was to get the website up and running locally with Weblogic. You were given a wiki page, but it had stuff like "Q will set up your DB access". So you had to figure out what Q is (a person). There was another part that didn't work at all. You had to ask around until you found your way to a guy named Venu, who could fix it. It was like a treasure hunt adventure game. But it was a good way to get to know the group..

I completed it in 1.5 days and everyone was amazed - 7 years of prior weblogic experience helped a little. Apparently an older iteration of the wiki page took two weeks on average. If a dev got it working in under a week - the department got excited they might have a rockstar on their hands.

With the new node stack, we made a cookbook. Now it takes maybe 5 minutes to get the app up and running on a brand new computer - most of that time is downloading and installing git and node.