Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
gangtao
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
61.
▲
by
gangtao
1y ago
I found a simple way to be success, rewrite some exisinting project using rust!
62.
▲
by
gangtao
1y ago
Baidu did this in the past and quickly lost his credit and market in China.
63.
▲
by
gangtao
1y ago
another UI library? But why?
64.
▲
Complex Event Processing Made Easy with Streaming SQL and UDF
(timeplus.com)
2 points
by
gangtao
1y ago
|
0 comments
65.
▲
by
gangtao
1y ago
I feel lots of search can be better replaced by LLM Just like this one The Hellfire missile travels at approximately 950 mph (1,530 km/h or 425 m/s). This is roughly Mach 1.3, making it a supersonic missile. The Hellfire is an air
66.
▲
by
gangtao
1y ago
My feeling, work with LLM is like teaching kids Dont say not to do somthing, always say do something instead This positive instruction style works because it gives the LLM a clear target to aim for rather than a boundary to avoid. It's
67.
▲
by
gangtao
1y ago
we use Docusaurus and netlify
68.
▲
by
gangtao
1y ago
you might check 1. Multi-Entity Structure Many Indian SaaS companies establish subsidiaries in business-friendly jurisdictions like Singapore, Dubai (DIFC/ADGM), or Delaware. The Indian entity provides services to the foreign subsidiar
69.
▲
by
gangtao
1y ago
there are some limitations as I know: 1. Insert Performance Degradation Users frequently complain that materialized views significantly slow down insert performance, especially when having multiple MVs on a single table. 2. Streaming Data P
70.
▲
by
gangtao
1y ago
Who would've thought that 50 years of 'SELECT * FROM reality' might beat the latest semantic embedding wizardry?
71.
▲
by
gangtao
1y ago
you can check https://github.com/timeplus-io/proton which provides streaming processing pipeline.
72.
▲
by
gangtao
1y ago
chose the one I am familiar with
73.
▲
Visualizing the UUID for Different Versions
(gangtao.github.io)
2 points
by
gangtao
1y ago
|
0 comments
74.
▲
by
gangtao
1y ago
just feel uncomfortale to need learn a new languages
75.
▲
by
gangtao
1y ago
looks nice! I will try this one
76.
▲
by
gangtao
1y ago
what is GDScript, why you use this language?
77.
▲
by
gangtao
1y ago
there are lots of TTS models, while what I care most is how can I easily use it, like clone my voice?
78.
▲
by
gangtao
1y ago
1. realtime features, timeplus can help build realtime features with realtime data processing capabilities 2. realtime data context, for AI LLM that need fresh data context to do a better job, timeplus can be used to process and service tho
79.
▲
Streaming SQL for Real-Time AI and Machine Learning
(timeplus.com)
1 points
by
gangtao
1y ago
|
3 comments
80.
▲
by
gangtao
1y ago
so why not to use C++?
81.
▲
by
gangtao
1y ago
I like this idea, you can use it to build something quickly
82.
▲
by
gangtao
1y ago
Thanks for sharing, ocaml is an interesting language which I think has it's pros and cons Pros: 1. High Performance with solid safety guarantees 2. Greatly improve developer productivity through advanced type features 3. Exceptional re
83.
▲
by
gangtao
1y ago
I also just learn it, glad you are looking for it
84.
▲
by
gangtao
1y ago
yes, I think so, you can change the tone how to response if you have your own prompt
85.
▲
An efficient and user-friendly hybrid cache lib in Rust
(github.com)
4 points
by
gangtao
1y ago
|
2 comments
86.
▲
by
gangtao
1y ago
the problems that can solve other problems, we call it meta-problem :)
87.
▲
by
gangtao
1y ago
Bambu Lab A1 Mini ($299-399) Excellent print quality out of the box Automatic bed leveling and calibration Very user-friendly with great software Compact size, perfect for beginners Creality K1 Max (~$599) 300x300x300mm build volume Fast pr
88.
▲
by
gangtao
1y ago
Post it on some social media like X , linkedin, or Bluesky The Cloudflare team might contact you soon! They might not care HN, which I dont know
89.
▲
by
gangtao
1y ago
This is a nuanced question that depends heavily on the specific criticism and context involved. Some criticisms of AI writing can indeed reflect problematic biases: Potentially elitist aspects: Dismissing AI use often assumes everyone has e
90.
▲
by
gangtao
1y ago
Micheal, the author of the project did some performance test and he might annouce it later, but one of his first test show Async Insert with Rate: 3422488 rows/second, but I beleive after optimization and tuning, it could be much faste
More ›