351–360 of 534 posts
Earlier quoted context omitted.
Simplicity is a driver for better abstractions. But now with AI, will we even develop new abstractions?
I was in charge of cleaning up a slop codebase by someone who has barely even heard of 'coding' before. Let's just say, it was abstract .
Earlier quoted context omitted.
A good interviewer won’t be looking for a single solution to the problem. I’d expect them to entertain the Google Sheets answer - it’s good signal that the candidate will consider what already exists in the world. I’d rather extend the problem: the team is spending considerable time iterating with manual entry, what would you do?
Complete agreement. "Excellent answer, that is what I would do as well, now what if we wanted to build it in-house?"
"Well I would probably go home and work on my resume because that's a fool's errand."
I hate going to work and reinventing wheels all day because the company I work for thinks it's so special that every business function needs a 100% tailored solution to solved problems. I much prefer working somewhere that's able to tailor business processes to conform to existing standards.
But maybe that's just me.
Earlier quoted context omitted.
I went to law school and a few of us students were engineers. For our first set of essay exams, the professors all instructed us to "just answer the legal question" and not include extra analysis. After the exam, many of the engineers didn't do well because the professors *actually* wanted you to weave the whole sylabus into your answer (i.e., discuss hypotheticals that were not actually part of the question asked),…
This is also what I learned the hard way. In many situations the customer doesn’t say what they really want. There are a lot of reasons why. I usually have to write down a lot of hypotheticals. If X is the primary concern, we should do Y. If U is the issue, we should do V.
Earlier quoted context omitted.
This seems like a perfect use case for a local model. But I've found in practice that the system requirements for agents are much higher than for models that can handle simple refactoring tasks. Once tool use context is factored in, there is very little room for models that perform decently.
What I hope to do with refactoring is to distill namespace and common patterns into a DSL. I am very curious about what tradeoffs you found.
Earlier quoted context omitted.
Complete agreement. "Excellent answer, that is what I would do as well, now what if we wanted to build it in-house?"
> "now what if we wanted to build it in-house?" "Well I would probably go home and work on my resume because that's a fool's errand." I hate going to work and reinventing wheels all day because the company I work for thinks it's so special that every business function needs a 100% tailored solution to solved problems. I much prefer working somewhere that's able to tailor business processes to conform to existing stan…
I had an interview question. What would you do if two different people were emailing a spreadsheet back and forth to track something? I said I’d move them to google sheets. There was about five minutes of awkwardness after that as I was interviewing for software developer. I was supposed to talk about what kind of tool I’d build. I found it kind of eye opening but I’m still not sure what the right lesson to learn was…
In case you ask “how can simple solutions be hard to understand and test?”
Lets just say you use single line bash scripts with multiple pipes, loops and very niche cmds.
It will work, it will look like nightmare, it will be simple -> one line.
I'm currently building a full-blown OpenAPI toolchain at work, where the OpenAPI document itself is the AST. It contains passes for reference inlining, document merging, JSON Schema validation, C++ code generation and has further plans for data model bindings, HTML5 UI...
Why? Because I'm working on a new embedded system which has a data model so complex, it blew past 10k lines of OpenAPI specifications with no end in sight. I said "ain't no way we're implementing this by hand" and embarked on the mother of all yak shavings.
I want all of the boilerplate/glue code derived from a single source of truth: base C++ data classes, data model bindings, configuration management, change notifications, REST API, state replication for device twins and more. That way we can focus on the domain logic instead, which is already plenty complex on its own.
I'm not designing all of this to be simple to develop. I'm designing it so that it's simple for the developers. Even with the incomplete prototype I have currently, the team is already sold ("you mean I just write the REST API specification and it generates all of the C++ classes for me to inherit?"). The roadmap of features for that toolchain is defined, clear and purposeful: to delete mountains of menial, bug-prone source code before it is ever written by hand.
Sometimes, it takes complexity to deliver simplicity. The trick is to nail the abstractions in-between.
Earlier quoted context omitted.
So my cofounder was talking to Stripe about an acquihire (this was after I’d left.) As part of it, he had to do a systems design interview. He got the prompt, asked questions about throughput requirements (etc.), and said, “okay, I’d put it all in Postgres.” He was correct! Postgres could more than handle the load. He gets a call from Patrick Collison saying that he failed the interview and asking what happened. He e…
I feel like if that's the thought process, that should be stated up front There's a ton of incredibly talented neurodivergent people in our ecosystem who would trip up on that question just because of how it's framed Because how is the interviewee to know if you're testing for the technically sophisticated answer no one in their right mind would ever write or the pragmatic one?
If anything, it's neurodivergent interviewers. If I insisted on a different design I'd either ask a question that's not solved by "just use postgres" or follow up with "ok, that would work, but what if ". Just failing a candidate for a correct answer is a prime example of why interviewing is so bad.