Well, I have written extensively on these issues already, so to summarize. Throttle: forces more actual answers, limits off topic comment debate Merge: makes messages more powerful, keeps sites on topic
@TravisJ So a chilling effect for anyone who posts a comment on an HNQ question, since their comment will inevitably be deleted as a result of everyone else getting into a debate and mods just mass-deleting comments? Results are never as clear as they seem. So now we have to apply a weight calculation based on popularity, making things even more confusing and complex.
@TravisJ I find those are never very accurate or useful.
Especially on small sites where the majority of answers are downvoted, yet the occasional upvote by some ignoramus on an HNQ answer results in resetting the counter.
one other point "against" a new site is "who's gonna pay for all this internet traffic and computer hardware" which I don't think is an absolutely insurmountable expense that a crowdfund or a couple well off developers can't put together
If someone was genuinely serious about a new site and it looked like it had a real chance to gain a critical mass of users, I would be willing to host it for free (at a DC, not out of my house). It'd be Russian or Romanian so the relevant laws would be important, but it would be pretty fast.
@djsmiley2k-CoW elected users of course, somebody people already trust on SE at first, and then new users who join for the first time who rise up to the level of expectations one would have from a worthy moderator
I've managed everything from programming / hacking communities to adult communities, and honestly it's all the same, modulo some interest-specific cultures and corresponding norms (and resulting rules).
@KevinB It optimizes PNG images, sometimes reducing size by 50% or more (occasionally 80%, usually more like 10-20%). It's an entirely lossless optimization, too.
The problem is that it uses a lot of CPU time to process a single image, so it won't be useful for user-submitted images (e.g. image hosts or image boards) without risking CPU over use.
Because it involves doing brute-force adjustments to low-level compression parameters.
It uses the normal PNG format, but it decodes and reencodes it using a variety of different parameters until it finds the optimal combination. The only way to do that is with a brute force test. It can't really be done at encode time unless you're happy with it taking 30 seconds to encode a PNG.
I'd be fine if it was a background service that does this while I'm away from my desktop, but then it changes the file md5 hash and if I wanna look it up years later I'm out of luck lol
@FélixGagnon-Grenier Think of it like H.264 (and H.265 I guess) parameters. It takes someone very familiar with video encoding to find optimal encoding parameters that minimize size without affecting quality. For images, each test encoding is fast enough that it can be done with brute force.
@DavidA It already is. Netflix and the like use it. Really all that's left to do is wait for GPUs with hardware-accelerated H.265 encoding being common.
@user1306322 Pretty simple. Just strip everything past the end of the image data.
But then you've opened a steganography can of worms.
4chan used to encode various illegal pornographic material in images of sinks that way.
[EDIT #1 by OP: Turns out this question is quite well answered by exiftool creator/maintainer Phil Harvey in a duplicate thread on the ExifTool Forum]
[EDIT #2 by OP: From ExifTool FAQ: ExifTool is not guaranteed to remove metadata completely from a file when attempting to delete all metadata. S...
it's plainly there if you open it with any image viewer though, I thought steganography was supposed to be hidden from the eye to a reasonable extent :p
I suppose "glitch art" or "pixel sort" images can hide it from those who are familiar with this technique, it wouldn't be unexpected for the art style to include such lines...
I hope I didn't just ruin both subreddits for the person using them for these purposes lol
Sometimes thumbnails are embedded in the image files themselves. I remember a case of one woman who took a NSFW pic of herself, cropped it to be SFW, but the original thumbnail was retained.
@DavidA Yeah but no one uses those anymore. They're legacy from the days when hardware was so slow that JPEG decoders took real time, and it would be inconceivable to decode each and every full-size image just to display a list of thumbnails. This was the days of Algorithmic vs Huffman coding.
(Algorithmic is superior but isn't often used because it used to take longer to decode. Also patents.)
Back in the days of dialup you could have a long image, with a surprise on the bottom. Since it took a couple of minutes to download a long image people would go make a coffee and return to their surprise.
On the downside (or upside, depending on your point of view), such patented or legacy functionality is often retained in image decoders because they need to adhere to the standards, and that results in a lot of really juicy security vulnerabilities, especially since the code is written once and forgotten.
GIFs (even the animated ones) are literally 30 years old. Kids these days are posting memes with them instead of something better and it kinda makes me sad.
@DavidA That seems like it'll be very useful for finding bugs in emulators. :P
(Which are a drastically underlooked exploitation vector: See scarybeastsecurity's stuff)
Heh, reading that more I see that it's pointing out the same problem of different decoders reporting different instructions for the same data. It was mentioned in the talk about interesting anti-RE techniques that could be done by fooling the decoder used by malware analyzers.
The one thing that really stands out to me in that is just how horrible libbfd is doing.
I'm currently the only HR worker at a medium sized engineering firm in Canada. There should normally be 2 additionally HR workers on-staff more senior than me, but one is on parental leave and the other position isn't currently filled, and I'm in over my head.
We have a senior engineer, "Francis...
Basically, Arduino was founded by five Italians, each of whom already owned electronics companies themselves.
They had come together, and in the late 2000s upon finishing their work, had all together registered the "Arduino" trademark in the U.S.
But in 2010, they tried to register the trademark around the world, but couldn't because it was already registered in Italy. It was later discovered that one of the co-founders had secretly registered it in Italy without letting anyone know.
The co-founder in question assured them that "it was all just to protect their collective investment" and his company (Smart Projects) still continued to pay royalties to the collective organization.
Later, in 2014, the co-founder in question left his company, and its new CEO essentially broke off ties with the collective Arduino organization and refused to pay royalties. He also renamed the company "Arduino" and created a website mimicking the name, which didn't mention the original founders at all.
This led to a weird situation for two years: the original Arduino organization could only sell products under the "Arduino" name in the U.S. (where they held the rights), and had to sell products under a different name (they chose "Genuino") elsewhere around the world (since "Arduino" formerly "Smart Projects" held the rights).
The original Arduino organization could do nothing against it as they had no legal standing.
That is, until, it later emerged that Smart Projects was selling their products in the U.S. and thus violating their trademark. They were sued, and eventually they settled with both firms agreeing to merge.