eredengrin

It depends on the language. I linked a set of c++ guidelines and for c++, they are correct: it is undefined behaviour to do signed integer overflow. Some languages do specify it, eg rust, and even in c++ it might appear to work, but even then it is still undefined and should be strongly avoided.

It's not that uncommon of a convention to strictly use signed numbers unless doing bit manipulation, eg the Google C++ Style Guide.

That's depressing and yet unfortunately not very surprising to read.

> but in practice, we just saw a fleet of major Matrix tenders go to system integrators who then apologetically cut Element out, despite AGPL - putting Element's funding for 2025 directly at risk.

I appreciate the transparency and you totally don't have to but not sure I fully understand this sentence, do you mind explaining that further?

> I feel pretty rugpulled here

How? Synapse is still being developed, this is just a commercial addon for huge customers.

> and not hijacking matrix.org community channels to promote it. They don't seem to be able to keep their hats and chairs properly separated

matrix.org has never excluded mentioning commercial entities, past blogs have frequently included information about etke or beeper for example.

> Linux distributions are obviously massive sprawling complex things

Just my perspective but I think this was also perhaps a product of the time when Red Hat really started growing, back when linux was a little more rough around the edges. These days it seems linux has enough polish that some (many?) corporations are happy to just use off the shelf ubuntu and call it a day and hope nothing breaks too badly.

[0] has many more details but CentOS was never a drop in replacement and the new CentOS model is a big improvement for anyone wanting drop in replacements. I agree with everything else you said though, they were independent of IBM for nearly 3 decades and always were and still remain among the most active linux contributors (just see any of LWN's development statistics articles).

0: https://medium.com/@gordon.messmer/in-favor-of-centos-stream...

> A big part of why Nintendo prevailed in their lawsuit against Yuzu is that they proved the emulator could not be used without extracting encryption keys and bypassing copy protection.

Not a lawyer but as I understand it, the case resulted in a settlement and as such no legal determination was made. They didn't prove anything in court and no precedent was set regarding the legality of emulation.

With so many comments on the state of the job market, I'm surprised to not see even one mention of career fairs. It's quite common for universities to host these and while they have their own set of disadvantages, it seems they could also provide a lot of value in the current market. The physical presence would ensure that it's a real person and it makes cheating significantly harder. Deciding a time and place would allow lots of companies and applicants to meet within a few day period so it would be pretty efficient. A larger sized fair might last for a week or two and most companies would stay the entire time while applicants would just buy a one or two day pass. They could potentially even do in person interviews within the same period which would further save on hiring costs. Have a few of these in each region per year, it seems like it would be doable, it's just a matter of getting everyone on board (of course, easier said than done).

You could maybe even use the fair as a screening to give applicants a boost in future online applications - if they seem like a good applicant after talking in person but perhaps not the exact fit needed for current open positions, just flag their career account internally as a verified high quality applicant.

Short of career fairs, verifying identity and employment history might be valuable and it seems like LinkedIn or some competitor should be able to do this. If a company can verify itself through a reliable process and then publicly mark accounts of employees who have been employed there for whatever duration, that seems like a low hanging fruit. In fact it sounds so obvious that maybe there's a reason they haven't done this yet? Any reason someone could think of for why this isn't already happening?

> An app that syncs your files just needs access to those file, not the entire filesystem - which is what MANAGE_EXTERNAL_STORAGE gives it. Scoped storage and the SAF would allow syncthing to do exactly what it needs to do. The problem is that the app maintainer is unwilling to do the work to bring this app into the modern era.

If I want to back up my entire filesystem, accessing all my files is not too much permission. If I don't want to back up my entire filesystem and I don't want to allow an app all those permissions, there is a very simple solution: don't install the app. There's also a more complicated workaround: use GrapheneOS and set up storage scopes for the app. Now instead of these perfectly functional options that already existed, the choice has been removed.

> But in this case I agree with them.

I am not an android dev but I'm reading in this thread that there are other apps like file managers that still get this permission approved? If google were removing the permission entirely I could see an argument from an api maintenance burden perspective, but if it's a permission they're still granting to some apps then I simply have to disagree.

>> Android really needs to just allow direct file access to any file which is under a user selected folder.

> It does.

I don't think it does for anything under Android/data/ directory though does it? Which might sound like nitpicking except that there are some apps which only store their data under Android/data with no option to put it elsewhere, which means that there is no way for syncthing to access that data.

> This attitude always utterly frustrates me. I simply don’t understand it. Is your view that a company’s worth should be determined by how quickly it copies any random feature introduced by a ‘competitor’?

Yes I let my frustration and cynicism obscure my message, my apologies. To be clear, I love that apple copied this. I think they should've copied it as soon as GrapheneOS did it, if not invented it themselves. One of the most frustrating things to me about apple is that they claim to be the "pro-privacy" company, but then in many ways they are worse than what you can find in open source projects that in comparison could be described as hobby projects (with all due respect to the GrapheneOS devs - it's an excellent project and does actually get a fair amount of donations afaik but at the end of the day we're talking like maybe a few full time devs vs apple's workforce, it should not be hard for apple to compete).

It's just tough to believe that apple is so resource starved that they couldn't do better than this. I want them to be pro privacy, I want them to put pressure on google so that everyone can have privacy regardless of their OS. Others in this thread have mentioned great ideas about implementing these sort of permissions at the contact field level, I love that idea and I hope apple does it too, although I'm not holding my breath. I can understand the use-case for WhatsApp wanting my contact's phone numbers, but there's no way it needs the addresses/birthdays/anniversaries or anything else I have on my contact cards, and even if it did come up with features related to those fields, it should be my choice whether the fields are exposed or not.

> Do you understand that if every person that voices these complaints had things ‘their way’, then Apple’s collective development backlog would be filled to the brim with stuff that’s utterly original?

Sure, there's a lot I'd love for apple to do and I fully understand I'm just not their target audience. My original complaint was specifically directed towards the mismatch between their marketing and the reality of apple products when I fully believe they have the resources to align those two things. I just did a horrible job communicating that.

> And, if Apple had done it as soon as GrapheneOS did, can you honestly say that you wouldn’t still be complaining about Apple copying things?

Again, not complaining about apple copying things. Please copy all the good things and then make them better! Copy them so well that it makes google want to copy them too, then one day if it is not practical to use GrapheneOS, maybe I won't mind using stock android/ios so much.

> What would actually satisfy you? All this OS culture war BS aside, is what you’re expecting of Apple here even remotely in line with your professional experience?

I probably can't ever be fully satisfied but if apple passed the (low) bar of allowing third party app stores/sideloading and had a decent open source app ecosystem (which in my understanding is pretty intertwined with the sideloading), I'd legitimately consider using an iphone, if only because I have nearly as many complaints with android/google. Based on what is happening with apple in EU, I don't see it happening any time soon but maybe in a decade or two if we're lucky they'll be beaten into compliance. No idea what your second question means.

In spite of my earlier comment, I do think it is worthwhile to stick around a little longer to see how it goes. As Arathorn mentioned, the latest ElementX release just came out and they are saying it is production ready. I got the update yesterday (I was using it in beta before) and I did notice some of my rooms syncing faster in the ui. Once SchildiNext rebases I am hoping it will be something I could finally recommend to people, at least if voice/video aren't needed. If voice/video is a requirement for you, then I could totally understand going to discord, although I am hoping that with the native matrix calls it could change that once the ux is equivalent.

I wouldn't go so far as to say the ux utterly sucks but it's certainly not something I've been able to recommend to most family and friends yet (I can't be too hard on matrix since I have to use Teams sometimes which is simply worse in pretty much every category I can think of, which is sad since the microsoft product teams replaced, skype for business, was itself light years behind teams).

It's a bit unfortunate to see matrix still not there yet, but on the other hand I'm still hopeful for the future. I think ElementX is the result of realizing that an excellent chat experience is what is needed. I also think the matrix foundation governing board elections [0] show that the community agrees that a rock solid chat experience is what is needed. At least for the individual members category, none of the candidates who were elected had a platform statement advocating for more work on the experimental matrix features. Sumner (who I think had the highest support by a good margin, although I can't find the exact numbers right now) specifically stated that experimental matrix features should not receive further investment until the baseline features are acceptable.

I realize that getting funding is a beast of its own and that is probably why some of the experimental things took focus at times so I can't fault that, but I really do want matrix to be something I can eventually recommend to all my family and friends instead of whatsapp or discord. I think it is slowly heading that way..

[0] https://matrix.org/blog/2024/06/election-results/

[1] https://matrix.org/foundation/governing-board-elections/#nom...

Sounds very relatable coming from c++. Managing dependencies is generally not too fun, it's very easy to waste a whole day (or more) fighting with the build system/package manager/linker. I like to think it's the tools but then you see a really smart c++ developer and that excuse starts to sound a little weak. It seems c and c++ attract a relatively high average percentage of developers who are willing to go learn the ins and outs of build systems and how compilers and libraries all interact.

I'm curious about the happy path of ocaml vs c++, I'm not sure how much of a happy path there even is in c++. I guess if every dependency uses the same build system as the main project, maybe that's a happy path? Hard to call cmake a happy path though, more so just a tolerable path. Conan sounds nice in theory but my experiences do not match the hopes I had. I remember golang dependencies not being too bad when I tried it, really need to mess around with rust a bit to see what that's like too since I hear so much good about it.

The way you phrase that makes it sound like you had a negative experience, was that your intention? For all we know he didn't even see it, there's thousands of issues and PRs and it's not like this is his full time job, so missing a message sounds like a pretty expected and neutral interaction to me. There's a reason they used the word "shockingly" to describe his response time, it's because it's amazing that Wez can respond to even a fraction of all the messages as quickly as he does.

Also in case you wanted to bump up the priority of that bug, he does have github sponsors enabled.

Wezterm has built in ssh support where if you open a wezterm ssh window to a remote host and then open new tabs/splits within that window, the new tabs/splits are added as part of the existing ssh session automatically. It also allows you to host the session on the remote machine (kind of like tmux) and when you connect you get back all your previous tab/split organizations. From that perspective it can make sense to have the terminal own the layout organization.

Wez left the issue open and said "I personally have no plans to implement this any time soon.", so I have to imagine he's open to PRs. He just has other things to spend his own time on.

Is that assuming windows are rolled down or do they just speak nice and loud? I'd consider most drivers in my area to be relatively easygoing in heavy traffic but this sounds fun to try nonetheless.

To give a concrete example of what others are saying and why I also really do not like discourse, the performance issues in combination with other bugs and limitations makes it extremely clunky for large threads. Go to something like [0] and try searching for a piece of text. In both Firefox and chrome, the discourse search box totally breaks for me: I can navigate to the first result, but after that future attempts to bring the search box up fail unless I go back to the original link and search again (ie remove the comment id suffix from url). The native browser search of course is not sufficient because 90% of the thread is not loaded, and the lack of proper pagination makes it hard to search with external engines. After a few months of frustration I finally learned about the hidden /print endpoint to dump the thread into a single page which helps, but still does not give me warm fuzzies about discourse.

I already prefer native browser search and proper pagination instead of loading the page on-demand, but if none of these bugs existed, I could at least tolerate the decisions. Given the bugs though, these design choices really grind my gears in a special way.

0: https://discourse.ubuntu.com/t/please-test-autoinstalls-for-...

Thanks for the perspective, I sometimes have completionist tendencies so it's hard to skim without feeling I'm missing parts, but also it can certainly lead to motivation running out at times. And thanks for Crafting Interpreters, I am really enjoying it so far!

Do you have any thoughts about how to read some of the next step books like Engineering a Compiler/Dragon Book? I don't normally read large technical books end to end so I'm curious how others approach it. I am going through Crafting Interpreters right now and I like how it has well defined checkpoints that help me know that I can move on when I understand the current material. I skimmed Engineering a Compiler and it looks like it has exercises as well, do you recommend all/some of those, or any other methods?

Also I did some searching to see past discussions of Engineering a Compiler and found this interesting comment thread from Bob Nystrom for anyone interested: https://news.ycombinator.com/item?id=21725581

This was the first thing the title made me think of and I second the recommendation. Watched it for the first time last year and it's the most fun I've had watching a movie in a long time. Soundtrack was also notably excellent, both music and sound effects.

Yeah would've made way more sense. Either that or centigrade.

Same. I sometimes use Edge when a site is broken on Firefox and I get into trouble there because it has super weird behavior when you highlight text. Very annoying.

Not sure what you're on about. The article itself states they're typically 1mm to 2.3cm in length, and most of them that I've ever seen are closer to the 2cm than 1mm. Probably helps that just about anything 1mm in size looks about the same from a normal viewing distance, so there's reason to believe most people would have seen more from the bigger end of the spectrum.

Regardless of the size I quite like them, many are quite pretty and fascinating to watch.

There's also sites like https://www.blankapparel.com/ that will sell you individual blanks from quite a variety of brands (including Bella+Canvas).

My experience with Bella+Canvas was that the ratio between the sleeves/shoulders and torso felt really disproportionate in a way that I haven't felt with most other shirts, so they don't really work for me. The material itself did feel nice though.

Okay, so? Are you implying they would make the bridges closed source if they could? Then why do they still maintain the open source bridges instead of forking or writing their own, and why do they sponsor devs to make new open source bridges instead of contracting them to create closed source ones, and why do they dump a bunch of money into the matrix foundation with no immediate benefit to their business? Not sure why it's so hard to believe that people might try to support themselves while improving the open source ecosystem.

Yes, the clients (both beeper mini and beeper cloud) are closed source. That's their business model - open source bridges that anyone can run if they wanted, then they just make it more convenient if you use their services by hosting it all for you and giving a nice polished client. The comparison was to libpurple devs - bridges are the equivalent of libpurple. This is like if libpurple devs decided to write a closed source client based on libpurple and then charge for it. Sounds good to me if it lets them keep working on the open source stuff.

Beeper Cloud has had google messages support since at least September.

> What’s weirder, to me anyways, is that you talk about open source as if it’s a staple value for you, yet here you are carrying water for commercial SaaS startup. Comparing their efforts to the likes of those who created Pidgin and libpurple.

Beeper has done a ton of open source work on matrix bridges, both themselves and through sponsoring other developers. I don't see how it's out of place to compare them to libpurple devs at all.