HNHacker News
TopNewBestAskShowJobs

lewiscollard

728 karma · joined August 28, 2015

Django by day, petrolhead by night, cat BFF all the time - root@lewiscollard.com - https://exhaust.lewiscollard.com/
submissionscomments
lewiscollard··on Goldman Sans
> There are many countries in the world. You're making an assumption that in exactly none of them could something crazy like this be enforced.

Licenses should not attempt to optimise for hypothetical legal systems that may exist somewhere and do not follow the laws of logic. "Put it in, just in case" could be used to justify literally any absurdity.

> and a tiny one for GS to include this language.

As a sibling comment by @wegs pointed out:

> [L]egal language like this is just absurdly poorly-written and stupid. Whatever you want legal language to say, you want it to be clear and unambiguous. That protects everyone involved[.]

Which is to say, Goldman Sachs is entirely welcome to spam their license with absurdities. But don't expect anyone else to come along for the ride.

lewiscollard··on Goldman Sans
> What’s idiotic about it?

It imagines a model of open source wherein I can take some Goldman Sachs code (the font), use it in my project, then start making legal demands of Goldman Sachs because I used their code. I would be quite surprised if such a license existed, because it would be a legal absurdity.

lewiscollard··on PinePhone: PostmarketOS Community Edition
> I think the point is that it has to excel over the iPhone/Android/laptop classes in some, significant, manner, or else it just gets sniffed at. This is an important thing to recognise in the quest for platform adoption.

If you see mainstream domination as the goal, e.g. to surpass Android or iOS, you'll probably be disappointed. I see it as giving those of us that care about privacy, security and openness a way to opt out of the Apple/Google almost-duopoly, just as Linux distributions do for the Apple/Microsoft one on the desktop.

For me, it only needs to be just about good enough to use as a daily driver, for now. I know I will be buying the shit out of one when these are available in July.

lewiscollard··on Major Linux Problems on the Desktop, 2020 edition
I don't have a use case. I am saying that it requires zero effort to treat filenames as a big bag of bytes containing whatever, encoded however anyone wants them to be.

Anyone arguing for case-insensitivity has to both a) demonstrate that this is a source of confusion in the real world, by actual people using computers in UIs that exist today, rather than "when I think about it that's kind of weird because it doesn't behave like Windows" or hypothetical oh-dot-texts and b) come up with a better plan that handles the world outside of US ASCII.

lewiscollard··on A one-year moratorium on police use of Rekognition
> Independent oversight? So .. like from a government? Cause that's the definition of independent, from the people for the people.

If you do truly hold the belief that governments are "the definition of independent", and that all governments act solely in the best interests of their own citizens...well, to borrow some words from Public Enemy: Can't do nuttin' for ya man.

We're literally talking about the government using facial recognition with no oversight and little public debate and no consent from the public, so you wouldn't have an excuse for coming out with this "government is all of us working together" lorem ipsum even if you were born 30 seconds before the article was posted and didn't know anything about the way that governments actually behave in practice.

You come so very close to getting it, in that you acknowledge that this isn't how governments work in reality, and then you're like "well neither is anything else though" as if that is somehow an argument for a strategy that you kinda-acknowledge cannot work.

And FWIW: I agree, it's fucked, it needs to stop, both in the public and private sector, and anyone that works on such tech should be shunned. But your assumptions are false to fact.

lewiscollard··on Two vulnerabilities in Zoom could lead to code execution
> I wonder. Are all the vulnerabilities and issues with Zoom because of its popularity?

In a sense, yes. I think Zoom has achieved far more popularity than Zoom was prepared for. Everything about Zoom feels like it was thrown together hastily and for a much smaller audience.

It may seem unrelated, but I think a microcosm of all this was: Attention tracking. They removed this feature because of its unpopularity after privacy people flagged it, which may have not happened if Zoom hadn't gotten popular. But it also shows that they have no internal brakes on this sort of thing; no strong product lead or anyone else able to push back on the PowerPoint prodders with "nope, not happening, that's fucking weird and creepy".

Can you expect good code to come out of that environment?

Notably, these vulnerabilities were in features I, a daily user of Zoom in these weird times, did not even know existed (code snippets and GIFs). Strikes me as more "shit rushed out of the door with no oversight and minimal code review to get boxes checked".

lewiscollard··on Zuckerberg: I just shared the following note with our employees
You are entirely right.

I've long said that optimising for "engagement", with no regard to what kind of "engagement" that is, is a paperclip maximiser.[0] Rewarding the posts that get the most comments regardless of whether they are positive or negative or thoughtful or shitposting, and the posts that get the most reactions regardless of whether the reactors (?) have or have not given it a few seconds of thought, is destructive to understanding each other and destructive to democracy.

See, I would rather Facebook not de-platform Trump, or any of his fans, or anyone else I don't like, but let's not pretend that Zuckerberg gives a shit about peace, love, and understanding. They have taken conscious decisions to prioritise "engagement" over anything else, because that is what makes them money. I get it, that's business, that is what they have to do. Likewise, we should do what we have to do to protect ourselves and to protect democracy: Zuckerberg and anyone that works at or has worked at Facebook should, to borrow a phrase, be unto us as heathens and publicans.

(You can extend this to Twitter too. I know, they scored some imaginary Internet points with left-leaning crowds by "fact-checking" Trump, but the truth is, with the algorithmic timeline providing exactly the same rewards to shitposting as Facebook does, they are as much to blame for this too.)

[0] https://wiki.lesswrong.com/wiki/Paperclip_maximizer

lewiscollard··on Major Linux Problems on the Desktop, 2020 edition
I would recommend Torvalds on this: https://lwn.net/ml/linux-fsdevel/CAHk-=wg2JvjXfdZ8K5Tv3vm6+b...

As for my own thoughts on top...

In US-ASCII, case insensitivity is easy. Outside of that, your operating system kernel has to have opinions on "Ö" being the upper-case "ö" and so on. But why stop there? Should "Ö" be considered the same thing as "O"? How about Ø? Or O with diaeresis[1], which looks exactly the same as "Ö" but is in fact a different character? How about Cyrillic O vs Latin O?

But that also means that your kernel now has to have opinions about filename encoding. Maybe someone will come up with an idea so obviously superior to UTF-8 that it'll supersede it just as UTF-8 ate everything else, or maybe nobody will. What I am glad about, is that filename encoding wasn't baked into Unix in the 1970s, or indeed at any time prior to about 1993.

So, your operating system needs knowledge of all of Unicode and to care about filename encoding for...what gain? It's only weird if your previous experiences with operating systems tell you that it is weird, and calling it weird is not the same thing as being an actual usability problem.

In fact, as witnessed by the fact we're capitalising the first letters of our sentences in our comments, I would argue that case sensitivity is the default in normal writing (the word "it" being a fine example; if I had said "IT" you would have read the sentence differently), and so if anything's weird and counter-intuitive it's treating case as irrelevant.

[1] https://en.wikipedia.org/wiki/O_with_diaeresis_(Cyrillic)

lewiscollard··on Words that don't translate into English
From my background (I grew up in London), "geezer" seems like a decent approximation to me, though there's no female version of that I can think of. I gather it means "old man" elsewhere.
lewiscollard··on Blur Tools for Signal
For the case of text, blur can be brute forced.

If you redact, say, a credit card number with a blur, and I know what typeface the number would have been written in, and have a reasonable guess as to your blur radius, it might not be infeasible to compare the blurred version of every possible credit card number.

If you redact an email address with a blur, brute-forcing every possible email address will be harder. But if someone (say) leaks information to you, and you merely blur out their address, it's not infeasible that someone else could apply the same blur to a known suspect's email to verify whether it was them or not.

Of course, with a large enough blur radius it's not an issue. Still, a non-zero amount of times, it's been done badly enough I've been able to mostly "reverse" a blur by just squinting and sitting back a few feet.

Always redact text with solid blocks.

I don't know how feasible this approach would be to human faces. I think Signal has blurred it such to make such an attack infeasible.

I also don't think it's sufficient; if you don't want someone to be identified don't take photos of them, full stop and/or period. Take the photo at the top of the blog post. Who on that day, had that a backpack with that type of strap, a blue mask in exactly that shade of blue, that haircut, and that exact BLM t-shirt, in that place at that time of day? That could be sufficient information for a "fingerprint", though maybe not deanonymisation.

lewiscollard··on Major Linux Problems on the Desktop, 2020 edition
The author may be thinking of this work from last year: https://lwn.net/Articles/784041/ I do not know if this has since been merged.
lewiscollard··on Major Linux Problems on the Desktop, 2020 edition
> All native Linux filesystems (except ext4) are case sensitive about filenames which utterly confuses most users. This wonderful peculiarity doesn't have any sensible rationale.

There is in fact a solid rationale behind this.

> No polish, no consistency and no HIG adherence (even KDE developers admit it).

See also: every operating system. It's unfair to pick on Linux for this since pretty much every Windows and Mac program seems to want to invent its widgets & unique behaviours, and this has been the case since at least the late 1990s when I got into computers.

lewiscollard··on Defold game engine source now available and free to use for commercial games
> OSI decided to appropriate the term "Open source" to refer to something slightly weaker than "Free Software" but much stronger than "you can look at the source code". They are not in their right to do that.

This is incorrect. Over 20 years ago people who would later be part of OSI _coined_, not co-opted, the term (within the scope of software). It was coined to mean _exactly the same thing_ as "Free Software"; the OSD was directly derived from the Debian Free Software Guidelines and there have only been two licenses, as I recall, that are OSI approved and the FSF have said are non-free.

You also confuse "OSI-approved license" with "GPL", because the OSI (and FSF) have approved permissive BSD and MIT and many other non-copyleft licenses. Here you go: https://opensource.org/licenses/alphabetical

lewiscollard··on It’s Time to Get Back to RSS
> and when they moved to pay only (or they closed? i can't remember)

The incident you are thinking about, would be when they said they were going to close the service entirely, because of the volume of signups they had in the aftermath of Google Reader closing. https://blog.theoldreader.com/post/56798895350/desperate-tim...

In the end they did not do that, but realising any service I sign up for might close just as readily as Google Reader put me off RSS for years. (I am a paid The Old Reader user now.)

lewiscollard··on Show HN: Make Your PDF Look Scanned
Hey, I know I was not as positive & encouraging as I should have been initially, hence the edit on the end. But thank you for the kind words mate, that actually means a lot to me. <3
lewiscollard··on Show HN: SHA-256 Animation
Nice work, and that is a very cool approach to learning new things!
lewiscollard··on Show HN: Make Your PDF Look Scanned
Hey thanks - somehow, I entirely missed this being a Thing in the stdlib in 3. :) I've used it in my PR on baicunko's repo.
lewiscollard··on Show HN: Make Your PDF Look Scanned
It was while reading my above comments that I realised I should have shut up and contributed code instead, because that's definitely more helpful than being critical on HN, especially to a newcomer & their first project.

So that is what I've decided to do! First step: a PR coming out of getting this up and running on my Ubuntu box. :)

lewiscollard··on Show HN: Make Your PDF Look Scanned
Rate limiting (if by that you mean at the firewall or the web server) is not the way to do it. That shifts the problem somewhere else in the stack, into a place that isn't under version control in the same repository.

Consider: If you moved this on to another server, would you remember to enable rate limiting there? If someone else uses your code, will they know to enable rate limiting?

Rate limiting isn't a bad idea, but your security should not depend on it, especially as you have a way of securing it in your application. base64.b16encode(os.urandom(8)) will give you a 64-bit, filename-safe, as-close-to-random-as-reasonable suffix that should be long enough to make it brute-force-proof :)

The same reasoning applies to the cron job (I presume) that is cleaning your files - that's something you have to remember to set up for future (re-)deployments.

Edit: I'd also like to add that showing your code on HN takes bravery and this is, in fact, a neat tool that solves a problem I really wish didn't exist. So, good work on both counts :)

lewiscollard··on Show HN: Make Your PDF Look Scanned
It's still far short of being suitable for use of any private documents.

https://github.com/baicunko/scanyourpdf/blob/master/pdfwebsi...

This is rather less than secure; output files are named, e.g., "Scan_2020512_{four random lower-case letters}.pdf" into a web-server-readable directory.

That gives a total of 456976 different possible filenames on a day. It's more than feasible to brute-force that many filenames in the hour before files get deleted.

OP: I don't think randomly-suffixed file names are an inherently bad way to approach this. But you should definitely consider using a longer random string, and definitely consider not using the `random` module too (it is not secure and is not intended to be).

lewiscollard··on Inkscape 1.0
That is definitely abnormal. For me (on an i5-3350P (!) Linux junk desktop from Hell) it starts up in about one second, so eomething is definitely wrong (or you're using an even worse machine than I am, in which case something _else_ is definitely wrong). Maybe one of the Inkscape support channels could help you here?
lewiscollard··on New 13-inch MacBook Pro
"Market dominance" might overstate the case.

Apple makes up for a small percentage of all laptops sold, even if you compare manufacturers (rather than "Apple" vs "Windows machine"). They have a smaller share of laptop sales than HP, Dell, or Lenovo, and very close to the same (depending on when you look) as Acer and ASUS:

https://www.statista.com/statistics/818439/global-notebook-c...

lewiscollard··on Show HN: Verify JSON using minimal schema
This has little to do with dynamically typed languages, "JS" in the name notwithstanding. JSON is a serialisation format that is used in both statically & dynamically-typed languages. It makes as much sense to call XML validators bizarre as it does to pin this project as a weird byproduct of dynamically typed languages.

If your program produces or consumes JSON _and_ you think it would be a good idea to ensure said JSON is what you are/a client is expecting, then you need something like this in any implementation language.

lewiscollard··on Lesspass – open-source stateless password manager
You make some reasonable points, but two I find somewhat strange:

> Setting the master password to something insecure will undermine the security of all derived passwords

Then don't do that. The question is: Is LessPass in practice more likely to encourage this kind of poor security than its alternatives?

> Master key is related to the generated passwords. Using a strong PBKDF this should not be a problem, but why take the risk?

If its PBKDF is broken then so is its security. If crypto is incorrectly implemented in any other password manager then it too is broken. If the crypto algorithms used by TLS are broken then our online banking is broken. We don't avoid other things that rely on crypto just in case they might get broken some day.

[Mods: accidentally posted this as a top-level reply rather than a reply to this comment, sorry!]

lewiscollard··on United Kingdom to introduce security labelling on connected devices
Here's the original gov.uk press release, and as always (kudos to their team) it's beautifully legible:

https://www.gov.uk/government/news/government-to-strengthen-...

lewiscollard··on Tesla remotely removed autopilot features from used Tesla without notice
> Air bags, anti-lock brakes, and stability control are all controlled by complex computer systems and absolutely could kill you.

They are not complex in the way I was thinking. Functioning airbags and anti-lock brakes have been around since the 1980s. Analog sensors react to inputs, sometimes send it to a computer, which performs the right actions based on those simple inputs. Traction control has been around for almost as long.

I don't know for a fact that nobody has been killed by these simple systems malfunctioning, but I would be very surprised if anyone had.

I'll take any of your book recommendations, in return for mine: "Normal Accidents" by Charles Perrow. My takeaway from that book is: once you get past a level of system complexity that no _single person_ can reason about, system accidents are inevitable.

> Out of curiosity, what do you drive now?

1987 Mazda 323 GTX (4WD turbocharged hooligan edition of the 323/Familia). I am happy owning one of my dream cars :)

lewiscollard··on Tesla remotely removed autopilot features from used Tesla without notice
You are very perceptive with the "subscription model of commerce" thing. I do not like it either. But as for this:

> I can accept unintended bugs, this is a risk I would be willing to take (and hopefully neither me nor my family would suffer terribly for)

I will never buy a car that has computer-controlled systems complex enough that a software bug could kill me.

lewiscollard··on EU considers temporary ban on facial recognition in public spaces
As some that lives in a part of the country that voted for Brexit and actually knows a lot of people who voted Brexit: I can guarantee that nobody who voted for it did so because they wanted more privacy violations.

Your link is inserted in such a way that it looks like it should support the assertion that the security services supported Brexit, but it does not. Do you have a source for that?

lewiscollard··on Ask HN: A New Decade. Any Predictions?
I know some of these are obviously backed by winks, but I got a smile from this because they're only a hair away from being rather plausible. You should be writing sci-fi!

> Potemkin jobs abound in a world that doesn't require humans to work to produce anything but needs to keep them busy to stop them complaining about stuff.

"There is another theory which states that this has already happened." https://en.wikipedia.org/wiki/Bullshit_Jobs

lewiscollard··on Ask HN: A New Decade. Any Predictions?
- Firefox hits 20% market share on the desktop, prompted by Chrome's nerfing of extensions in Manifest v3 (e.g. uBlock Origin won't work on Chrome anymore, forcing nerds & privacy-conscious people to switch). Firefox on devices gets a tiny boost as a consequence. ETA: 2021-2022.

- Tesla implodes as soon as a major manufacturer releases a high-performance all-electric car affordable by people on modestly-good salaries that actually ships on time. Shame, because the Cybertruck is badass. ETA: 2023.

- In other EV news: Electric vehicles will dominate in unrestricted motorsports (drift racing and some classes of drag racing come to mind). ETA: Rule changes to allow EVs will come in 2021. A couple of big-name drivers switch that year. Domination is established in 2025.

- We get a real, viable open source Linux-on-phone with all the apps we need. There's that Purism thing already, but that's flagship money; this will be boosted by the Pinephone (and whatever follows) being cheap enough that almost every hacker will buy one (Android and iOS will continue to dominate the mainstream). ETA: 2021.

- Attacks in the Spectre/Meltdown genre (side-channel attacks on CPUs) get frighteningly good, and impossible to defend against. By the end of the decade we may need to either simplify processor design massively (and take the performance hit) or abandon the idea that processes can be effectively separated at all. ETA: new attack at least once every year of the decade. Complete rethink of processor design before 2030.

- A data breach on an unimaginable scale will happen. The Big One will be a tracking or analytics or advertising company. Equifax was bad, but this will be everyone's personal preferences, porn preferences, and (subset of) browsing history, which will easily be linked to identifiable people. ETA: any time now.

- House prices in the United Kingdom collapse. This causes global ripples similar to 2007-2008, in which we find out just how much of the financial system is betting on an asset price to keep going up, forever. Massive social unrest over the attendant bailouts. "That which cannot go on forever, won't". ETA: 2025. (I have hilariously mistimed this one in the past!)

← PreviousPage 4 of 7Next →