Google relies on writing syntactically correct code in a google doc with no autocomplete/auto indenting during their interviews. The cognitive load of this plus having to answer very difficult obscure algorithmic questions was too foreign from my normal workflow and too heavy of a cognitive load for me to be able to answer the question accurately in the time given. I hope they don't build production code this way.
Now I'm just imagining an april fools event where they replace the really nice internally hosted code editor with google docs, disable code search, and call it interview awareness day.
My wife worked there for the transition. Her very first .net page was the MySpace homepage (which was demoed by Steve Ballmer at CES, iirc). Microsoft was very unhappy with the decision to retain the .cf extension on web pages even after the transition to .net. There were some crazy levels of scaling happening behind the scenes—they had database requirements beyond the capabilities of sql server and there was code on the server side to route to different sql clusters based on the user's numeric id. They also had more page views than they could sell ads for. There was actually a pretty crazed level of new feature development happening at the same time—any time that friendster or some other site would release a new feature, Tom would insist that it be implemented on MySpace right away. At the same time, he was also requiring things like pixel-for-pixel identical output for existing pages/features through all of this.
That's called sharding and is pretty widely used nowadays. There is even readily available middleware that handles this for you, so no need to put it into the backend.
Keep in mind this is more than a decade ago. Sharding was almost certainly state-of-the-art and exotic back then (hell, it's arguably pretty unusual in the wild even today, even if a lot more mature of a concept).
Oh, of course! I did not wish to come of as arrogant. I just wanted to add the term (for easier googling if someone is interested) and a note that it's no longer necessary to modify your api to use that strategy.
I was at a CF conference with MySpace as a speaker and they talked about how often they went down but that they were switching to another CF engine with .net capabilities (BlueDragon). Their discussions and jokes about instability and going down due to bad code updates were really cringy and gave the impression that their platform was a mess.
Pretty certain we were at the same conference. There development practices horrified the group I was with in the audience. At a time that a new competitor appeared (Facebook) they made the fatal mistake of rewriting their software in dot Net. No new features for a year or more. Joel Spolsky wrote a pretty famous essay about the wisdom of doing that.
Performance was the feature. Friendster should have been social network of that era but they couldn’t scale and had to shutdown new user registration. Same thing happened to Twitter but they had no competitor in short form social networking.
I had an older friend that worked at CF when I was a teenager. He taught me a bit about relational databases and some other important stuff as he was one of the only developers I knew growing up.
I remember him telling me how I was wasting my time building in php/MySQL since CF was the future and serious companies wouldn't be looking for php developers. Glad I never pulled the trigger and bought their expensive IDE.
For proper early 2000s nostalgia, I reckon the backend should actually be Apache mod_perl, with the ".cfm" extension script aliased to Perl CGI scripts because management tell everybody "it's running on Adobe!"...
If you had enough rack space, mod_perl+Apache2 could scale to anything you needed. Hundreds of Gbps, hundreds of thousands of RPS. Inefficient, sure, but seriously flexible+performant and really easy to deploy. I miss the old stacks, as Nginx and the other httpds really suck in comparison (feature-wise) and Java is still such a huge PITA. I'd honestly rather use modern PHP than today's Java webapps.
Hilariously, ColdFusion strikes again. I should have used pound signs around the bad_memories variable. The above code would dump the string "bad_memories".
I don't believe that /> is needed anymore. You can just use >
I unfortunately work with coldfusion/cmfl on a daily basis at my company.
I am moving the servers over to Lucee from Coldfusion 10 and trying to have any new systems built on C#/.net.
The most difficult part is getting other people on board. My manager is fine with what I am doing. He saids he is too old to learn a new language so myself and the other programmer will have to support anything I built in .net
Let me tell you folks. Writing spaghetti code with no separation of concerns and no version control is not exclusive to php
> Let me tell you folks. Writing spaghetti code with no separation of concerns and no version control is not exclusive to php
Yowza. If I can give you some unsolicited advice: get some kind of version control in place! Even if it's just a local git repo at first.
There were two things that made working in ColdFusion almost tolerable. (1) The place I worked started using an MVC framework before I left. I was pleasantly surprised by how much it improved the experience. (2) Transitioning to CFScript instead of CFML tags made a big difference as well.
But at the end of the day, it's still ColdFusion, and it drove me bonkers.
The guy who invented that syntax was J.J. Allaire. Bill Gates passed on his advisors recommendation and didn't buy Allaire because he thought the price was too high. Instead he bought a company in Hawaii building what became ASP.net. However he was quite impressed with J.J.
But years later when J.J. started a second company, it was only a few weeks old when Gates swept in and bought it for millions of dollars.. Shut it down and moved everyone to Seattle.. He immediately assigned J.J. and his team to build a new project called Azure.
J.J. Allaire's other brother and partner in ColdFusion was Jeremy. Jeremy has started two successful companies since leaving Macromedia. They are BrightCove and Circle. I don't know about Circle but at one time parts of BrightCove were running on ColdFusion.
I interview for front-end at a big company (actually a different title, but effectively front-end engineers) and we explicitly do not ask trivia questions, which is what this guide focuses on. I might ask warm-up questions like "a customer says they click a button on the page and nothing happens, how do you investigate?" and that's it. The entire rest of the interview (1hr) is spent on a coding problem where we both walk through and discuss solutions, and I let them code and we talk about their code.
I truly believe that experienced engineers can learn 1000x more about someone's skillset in the first 5 minutes of them coding than any amount of trivia questions.
It really depends on the job - I'm interviewing jr. devs, and a lot of these questions are good for testing their basic knowledge. Not every front-end job is about cutting-edge javascript.
Yes, many electric cars have a buffer. Not all Tesla though. That's why it's not recommended to charge a Tesla to 100% everyday while it's fine with many other brands.
But Tesla did sell a few cars with different battery capacities and exactly the same hardware. If you got the smaller capacity then you got a buffer. Tesla could unlock the buffer remotely, if you give them more money or if you need to evacuate before an incoming hurricane or similar.
Yeah I think fair to charge the customer more to access more of the total battery capacity because Tesla is now assuming more warranty risk on the battery.
