Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
stitched2gethr
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
11 ms
·
91.
▲
by
stitched2gethr
4y ago
This argument comes up a lot but this is a very narrow case. Neither you nor I are building software for the super bowl, and if we were building something with such high stakes we should start way, way early and concentrate on core function
92.
▲
by
stitched2gethr
4y ago
> I mean, think about the software you use. Of all the shops that produce that software, which ones do you trust more than others? I'd be willing to be it's the ones that delivery predictably. At the very least the ones that de
93.
▲
by
stitched2gethr
4y ago
For me, TDD has been much more about making sure the code works the first time and to the spec and much less about the structure of it.
94.
▲
by
stitched2gethr
4y ago
Which ones are you contributing to?
95.
▲
by
stitched2gethr
4y ago
I believe this goes a bit further and decouples the submission from the review. My team reviews PRs very fast, but why should I wait 15 minutes between submitting and continuing my work. Or what about when I'm working late when no one
96.
▲
by
stitched2gethr
4y ago
Better to ask before the first interview, but that still takes more time than reading it off the posting.
97.
▲
by
stitched2gethr
4y ago
I have a different background than the author. I did take the time to learn the debugger, for multiple IDEs, and I've spent a *lot* of time in that rabbit hole. Not to discourage those using a debugger but it really was, for me, a hole
98.
▲
by
stitched2gethr
4y ago
First yes, but it's not difficult to imagine Uber has outgrown their first architecture.
99.
▲
by
stitched2gethr
4y ago
> People's expectations for what a website could offer rose tremendously, and would abandon a site if it wasn't up to snuff. Amazon.com is basic text and image elements on a white background. The full-feature sites often just l
100.
▲
by
stitched2gethr
4y ago
This is why I quit working at big companies. I'm in my second position at a startup and I couldn't be happier. - no constant meetings - no / less legacy code or dealing with systems that cannot be changed easily because they
101.
▲
by
stitched2gethr
4y ago
I use vimspector when I need a debugger, but find I don't reach for it nearly as often as I did in vscode. Nothing against those who are effective with it but I spent too much time crawling and not enough time solving. Print lines make
102.
▲
by
stitched2gethr
5y ago
What is it that you are looking for with this comment? If you already know texts.com is there a degree of comparison you cannot glean from reading beeper.com?
103.
▲
by
stitched2gethr
5y ago
This is how the Go community recommends you write Go code. It is pleasant.
104.
▲
by
stitched2gethr
5y ago
This is actually pretty impressive. More so than I expected, and I sincerely hope this opens the door to simple solutions for those who are still learning or don't code often. That said, this isn't the robot that replaces us, obvi
105.
▲
by
stitched2gethr
5y ago
Absolutely. The tech field is broad and deep. There are plenty of tech jobs which have nothing to do with data science. Data science is honestly a pretty small part of the tech landscape.
106.
▲
by
stitched2gethr
5y ago
I'm going to be honest, I'm a backend developer and I've never written a tree in production. I understand the structure, I've written them for practice, but for real world code it's just never come up. It seems that
107.
▲
by
stitched2gethr
5y ago
In practice I'm not sure there would be much duplication. Even for tools using the same libraries they are very often using different versions.
108.
▲
by
stitched2gethr
5y ago
I believe many highly successful products would contradict this claim.
109.
▲
by
stitched2gethr
5y ago
This is why the Solana model is succeeding.
110.
▲
by
stitched2gethr
5y ago
The Go team has been discussing converting some functions to generics _if_ they can maintain the compatibility guarantee. Without verifying, I believe this was discussed in Go Time ep 216.
111.
▲
by
stitched2gethr
5y ago
Very thorough. > In the face of ambiguity, refuse the temptation to guess. Great advice, but shouldn't it apply to the test traffic as well? Mimicking traffic patterns and making assumptions has a lot of downsides. You'll alw
112.
▲
by
stitched2gethr
5y ago
Seems like a good article but I got bored and distracted halfway through...
113.
▲
Speedscale (YC S20) Partners with New Relic
(speedscale.com)
2 points
by
stitched2gethr
5y ago
|
0 comments
114.
▲
CLI tool to inspect app traffic (L7 wireshark)
(github.com)
5 points
by
stitched2gethr
5y ago
|
0 comments
115.
▲
by
stitched2gethr
5y ago
I'm not convinced the distinction is that important.
116.
▲
by
stitched2gethr
5y ago
Vans shoes.
117.
▲
by
stitched2gethr
5y ago
hear hear!
118.
▲
by
stitched2gethr
5y ago
What is it about a large code base that requires an experienced developer to use an IDE?
119.
▲
by
stitched2gethr
5y ago
Total agreement. I love this planet but those things need to burn.
120.
▲
by
stitched2gethr
5y ago
Dude is jaded.
More ›