kernelsanderz

Really great project. I'm interested in how you've implemented the back-end? How did you productionize the GAN model? Or did you just pregenerate 1000 images? (ie. are the images generated on-demand?)

You can use a super-resolution tool like https://letsenhance.io/ to upscale the image which might make it better for printing.

Devices like the Pwnagotchi (https://pwnagotchi.ai) make this ridiculously easy and really fun. It’s getting very popular. Definitely recommend giving it a go if you have spare raspberry pi zero lying around.

This is very frustrating. From what I understand of the publishing industry, the Kindle sales are not counted towards the main NYT best-seller list, so all the focus on the initial launch is around selling the physical copy. This is said to be done to prevent gamification and manipulation, but this still goes on with the hard copies also, where would-be best-selling authors often pay third-party companies to buy a lot of copies of their books from certain bookstores.

Source: A friend of mine who has been trying to get onto the NYT best seller list.

I'd highly recommend watching this documentary called Paywall [1] which explains the unhealthy dynamic between academia and publishing. TL;DR - academic prestige and career progression depends on being published in highly rated journals.

[1] - https://www.youtube.com/watch?v=HM_nWsdbNvQ

I love that it shows memory bandwidth and also what generation/speed of PCIe slot is being used.

Thanks for the link. From what I can read it seems that there needs to be some concept of pausing or stopping and resuming a stream.

This is one of the outstanding issues with promises too. That there is no standard way to cancel them. I hope TC39 are considering this as part of the spec.

It's less of an issue in the client, because you are less likely to encounter big firehoses of data. But I have had the issue with webrtc data processing, and some flow control / back pressure mechanisms would be awesome. We were able to just use node streams, but would be nice if there was an elegant solution for async iterators too. Node streams are good, but also add a lot of complexity.

And debugging back-pressure issues can be very tricky in my experience, as they can manifest themselves in a variety of different ways.

The font list menu item doesn't seem to do anything when I click on it. Using safari on latest iOS

Hi. Seems like an interesting project. Perhaps would be more interesting to HN users if you described how the algorithm worked? Or even better, published the algorithm/model as open source so others can learn from it? I'm genuinely interested in how curation can work in the absence of traditional search engine signals like links or social engagement numbers.

Great library. I really can't wait until async iterators hit the language. I think they're much more elegant than node.js streams, but I'd like to understand better how critical concepts like back pressure are implemented with these constructs.

What a great idea. Have shared it with my family for my nieces and nephews!

That's so awesome. I guess the webrtc audio encoding libraries would need to be updated in the browsers to take advantage of these improvements though?

Given that Opus is built into the browser, is the format backwards compatible? Or will browsers need to update their Opus Libraries to get these changes?

I have the same problem. I basically use Dropbox as my work folder and put all my files in there. As a developer I have millions of small files. It takes 2-3 weeks to sync the dropbox from scratch (currently 285gb of dropbox files). And my CPU is regularly pegged at 100%. I basically have to switch Dropbox off when I'm not plugged into power on my Macbook Pro in order to conserve battery. I've contacted support several times, and their response was basically "don't".

I'm on one of their top plans (2TB). What's the point in having such large plans when the number of files you have causes the whole service to break down?

Been searching for a good alternative to Dropbox. Something that would be CPU efficient, safe, secure and reliable. I also use their packrat plan (saves a history of every file ever), which is nice too.

Any other HN readers have any alternatives? I really just want a plug and play solution, and not have to roll my own.

I don't trust Apple iCloud for one bit as they've been quite inept at running cloud services. I've tried Google drive with poor results too.

I would imagine contention ratios would also factor into pricing. At least in Australia you might have a 8:1 ratio of actual bandwidth available for a consumer plan, and 3:1 for a business plan. I'd imagine that you pretty much get all the bandwidth you pay for in a data centre. I've certainly saturated 100mbps connections on servers in the past, 24x7. But perhaps people more knowledgeable than me could comment on this?