antpls

Why didn't you profile the Python code and only rewrited the hot loops?

columnar layout is only faster if reads are sequential

I tried few years ago, and it didn't fit my needs. Just tried again :

  - must sign up to use the application
  - impossible to select several items to mark them as read/unread (possible in Aggregator, and I can't live without it)
  - the grouping is based on the feeds (same as Aggregator), but I'm looking for label grouping ala Gmail
  - must pay for the search/label feature
  - in text only mode, the items are hard to read. I like Aggregator's way of seeing the source's favicon next to the items
  - ads

RSS is free open standard, it's not supposed to change and does not require maintenance. Aggregator's last update was almost 3 years ago and the app still fully works as intended today. I'm not looking for an app that is updated every week for new features.

I tried all the free Android RSS readers under the sun, and none has the perfect combo of features I would like.

So far RSS is unmatched for me. I started with Netvibes when I was a teenager, then I stopped for some years, then I installed Aggregator on Android a to be able to follow RSS feeds again.

I'm looking for a smarter RSS reader on Android than the existing ones. Features I'm looking for :

  - automatic deduplication of items
  - smart labels powered by ML
  - My own grouping based custom filters AND smart labels
  - easy access to discussions that might exist on social website for a given link (twitter, HN, reddit)

That might be the usecase that would get me started with Android development, but I have not decided if a webapp would be better. Also every news businesses and news distributors on Earth (including Google) are trying to kill RSS to make more money, I don't see who could stop them.

I confirm some Reuters articles just disappear after opening the page. Been doing that since the new redesign, a few days ago. We are witnessing the slow death of Reuters.

From what I read about Chia, the marketing is the exact opposite : most people have unused storage space, and they will never use it (phones and computers). The Chia network proposes to make that free space more "useful".

How one does explain that the price of Ethereum follows the price of Bitcoin, while both networks are separated, and follow a different roadmap?

In this case, wouldn't the value be multi gigabit, not the key?

I was precisely replying to the parent who is wondering about a path to profitability for Discord.

In addition to other comments : advertising. Imagine having to watch, or listen to an ad before you can join a channel. Imagine contextual ads in-between text messages.

Today I browsed Google from my desktop PC about "Gundam" (the japonese anime) for about 1 hour. Pretty niche topic.

Then when I was done, I closed all browser tabs, I unlocked my smartphone to check Instagram, and the second ads I get is a Gundam ad. I have no idea how my shadow "desktop" ad profile and shadow "mobile" ad profile matched. I share no accounts between my desktop and my smartphone :

  - I use different Google accounts on my phone and desktop (although I occasionally log in one or the other in private tabs)
  - The internet connections are different : 4G for phone, fiber for desktop
  - On desktop, I use Chromium with Ad block Plus
  - I do not use facebook nor instagram on my desktop PC

So how the f. did Instagram know about my Gundam browsing on desktop.

This must stop.

Since Reuters stopped its RSS feed, I found an interesting alternative : Aljazeera English.

Aljazeera English has almost the same content than Reuters (like literally, exact same articles words by words), has RSS feed, and also has interesting news and viewpoint about Africa, Asia and Middle East.

AP News has no RSS feed.

Aljazeera English on the other hand has almost the same content than Reuters (like literally, exact same articles), has RSS feed, and also has interesting news about Africa, Asia and Middle East.

Closest thing to mind is Numba's cuda JIT compilation : https://numba.pydata.org/numba-doc/latest/cuda/index.html

Then you have Cudapy : https://github.com/oulgen/CudaPy

But in my opinion, the most future proof solutions are higher level frameworks like Numpy, Jax and Tensorflow. TensorFlow and Jax can JIT compile Python functions to GPU (tf.function).

In my opinion, no, it makes no sense to write ML in C++ :

- Python allows for higher level description of algorithms, which means researchers can focus more on the ML stuff and less on low level details.

- There is no performance gain in going from Python to C++, because in both cases the models are compiled to specific binary formats to be executed on dedicated hardwares. TensorFlow enables accelerators not only for training, but also for data transformations and preprocessing.

When the topic "Commission" is trending, it's always full of furry porny stuff, I have no idea why

How I understand it, please correct me if I'm wrong :

All transactions are visible, sure, but Bitcoin also enables mass dilution and mass money washing.

Let's say there is a bitcoin address A on which you have identified that 50% of the Bitcoins are legit, and 50% came from a suspicious flagged address B.

Now what? Do you also flag A as "suspicious". What if A also made thausands other payments, do you flag all the recipients as suspicious?

Let's say you are drastic and your external system only allows payments from the addresses with a 100% verifiable and clean history. Now all I have to do as an attacker is to make a payment to your address from a flagged address I own, and all your bitcoins are also now flagged for life.

You quickly end up flagging the whole address base as suspicious, because there is no way to :

  -  block deposit 
  -  differentiate Bitcoin X from Bitcoin Z

I guess something like smart contracts could change the rules, but that would be a completely new network and token

Or am I missing something ?

Telling people "just use flatpak and you are safe" is plain wrong and dangerous.

Containers are _not_ safe, that's why Amazon created firecracker, and Google created gVisor.

Containers are a _convenience_, not a security measure. All security precautions apply to container apps as any other apps.

Flatpak is backed by Linux containers, which are not designed to be secure against hacks. Flatpak (like Snap) is a convenience to distribute and install apps, not a security protection.

Technically it doesn't just prevent from selling SaaS. It prevents from selling SaaS without having the consent of the open source project first, which means negotiating a deal.

The whole point of WASM bytecode is that it doesn't need to be reviewed. The worse it can do is "stealing" your CPU time, but WASM was specifically designed to safely run third-party programs in the browser

As a EU citizen, I wonder how that is different than the China Communist Party methods.

How do you know it is the full data, and some of it was not removed ?

Same. The only useful addition was type hints.

Now, every two releases they add a "language feature". Some library maintainers jump in the bandwagon, and WOOSH! : half the ecosystem depends on CPython to run. Bye Jython, Cython, PyPy, Micropython, etc and bye the code analysis and code transformation tools : all outdated until more funds or voluntary dev are willing to fix them

In an ideal world, the community should fork Python 2.7, rename it and add type hints. But that's too late, all the dev time was spent on making libraries compatible only with some recent versions of CPython

True, because that was already a hard problem by itself. And it got even worst with the recent additions.

Sorry, I disagree.

1. All you need to do a remote whiteboard is a microphone and a mouse. This is a great opportunity to create new softwares that actually help collaboration and improve the design process rather than being stuck forever with a pen and a paper.

2. What you call "personal relationship" is, at the end of the day, employee relationship and politic to be promoted. You are in a long term position of power on your employee, which allows you to impose your point of view. Stop pretending to be benevolent. Most new hires do what you tell them to do, it doesn't change with work from home.

3. I think that's just because you are not a good manager (sorry), or your own managers are bad. If you actually cared, you would respect people time and have clear rules that show it. You can send a "have a good afternoon" at 6PM, no one would find that awkward

Doesn't Vitalik/the dev team own the "Ethereum" name, which means they decide what get called "Ethereum" ?

If there are two versions of the chain, because of a 51% attack or a fork, there always will be only one dev-approved chain called "Ethereum" on the exchanges, and that's what the users will buy.

Well, that was just one public blog post. I would expect Google's R&D to go beyond what they publicly present.

I am not sure why you are downvoted.

Google has been doing electronic with ML for some time now : https://ai.googleblog.com/2021/02/machine-learning-for-compu...

Python and Tensorflow work just fine for that task, the speed comes from LLVM and dedicated hardware such as TPU. The language in which the high level ML network/algorithm is specified doesn't matter much.

It is always good to have several paths explored, but I'm skeptical on the Julia hype since its beginning.

Wouldn't it be possible to create a new type of index in Postgres (or maybe it already exists) which would take all the data of a column and simply lays it out in columnar format in memory, with all the benefit of compression ?