766 karma · joined October 27, 2011
[ my public key: https://keybase.io/freewizard; my proof: https://keybase.io/freewizard/sigs/2Eoeh17HRRF0PO7Ea1OGwEtjmLb6A7mfaOmxyGCmPIk ]
Ops is good q, based on my observation, most of these implementations has a Dockerfile at best, CloudFormation/K8s is very rarely considered. And probably bc where this community was originated from, Amazon is pretty much suggested against.
- CJK support is tricky for almost all of these search products, but postgresql is probably the only one that needs additional extensions, and some extensions are not updated for a while.
- Scaling is tricky, bc on one hand, your search and other normal queries may not have same load pattern, but you only have one schema in multiple nodes, so sometimes you have to waste some resource; on the other hand, dependencies of the 3rd party CJK extensions makes it hard to use managed instances and maintaining your own is time consuming.
- Most other search engines have well tuned APIs for typical use cases like filters/typo/etc, but you have to build your own in PostgreSQL. Of course it’s not entirely bad thing bc PostgreSQL does have most flexible query capabilities for those less than typical use cases.
End of day, search for end users have many semi-edge cases, one have to try all these engines to find the best fit. For large use cases, operational cost and user tolerance of search errors are all part of consideration. A conventional wisdom is do less work in the early days , and find better solutions later if have to.
Gaps/opportunities for Fediverse community:
- sustainable mgmt structure and plan for large nodes
- features to import content while not breaking existing ecosystem (it's currently possible to move followers and export content only, but not import)
Not sure congress or FTC can do much. Ban click ads? Ban Chinese advertiser? Ban links to Chinese-owned websites? Ban BGP to China?
- 750k row of sample data is large enough for a leak by itself, many on reddit/twitter/fediverse have already started to explore the data set for gender ratio, age composition and frequency of raping cases, etc.
the 750k row sample file seems very real, with names/id/gender/education/marriage/mobile/address/crime record, the latest birthday is 2021; it also contains a table of police cases in Shanghai dated back to 2000s. it also has urls for id photos, plus photos taken in police stations and customs. Hopefully those urls are not actually accessible to the hackers.
the only suspicious part is that it's priced only 10btc, $$$ but still peanut compare to the value for data covering such a large population in China
I feel sorry for the progress of gender equality in Silicon Valley.
[1] https://www.businessinsider.com/spacex-paid-250000-to-a-flig...
Running external display at 4k@60Hz is possible but not straight forward, it requires patch core graphic framework, or using 3rd party boot loader. Newer models do not have this limitation afaik.
Didn't Google just said they allow Spotify users to choose between Google vs Spotify payment system? Why are they now requiring Bandcamp to "exclusively use Google Play Billing"?
Dune was not bad, but personally I love Dune II a lot. It was the first RTS I played, and it had such a huge influence on this genre for the years to come.
Typesense in my opinion is much better on performance (ymmv but in my case it's ~15% disk/ram, 5% of index time comparing to Meili). However, it's a bit behind on easy to use: the auto schema api has some limitation, also lack of official arm64 build (maybe it's a c++ vs rust toolchain thing, it takes quite some time to compile by oneself, with many c++ libs used from different vendors like google/baidu/etc, and some lib even requires SSE)
It's fun to see both come with comparison charts [1] [2], I dunno who started this but competition is good for the industry (except I'm mildly concerned GitHub stars are used in PR like this), and I hope both get better over time.
[1] https://docs.meilisearch.com/learn/what_is_meilisearch/compa...
[2] https://typesense.org/docs/overview/comparison-with-alternat...
For example, Netflix and Prime Video are almost equal on total number of shows (~5.5k movie + series) but on Netflix only 4% shows are from pre-2000, while Prime has 20%. Also HBO Max has half size of Netflix library, but when filter by IMDb rating and voters, the gap is much smaller, and in certain cases HBO even wins out.
In terms of language diversity, Netflix is doing the best with 45% content not in English, that is far more than anyone else (except Zee5 which is primarily Hindi). However when it comes down to a specific language, the number is usually much smaller and around 100, which is less than ideal for someone primarily consumes French/German/Chinese/Korean/etc. (tweeted more about this [2])
[1] https://www.movieofthenight.com/compareservices
[2] https://twitter.com/freewizard/status/1480246184887398409