Post ATH2WpuPvHr3TdwMMK by Agartha_Noble@amala.schwartzwelt.xyz
(DIR) More posts by Agartha_Noble@amala.schwartzwelt.xyz
(DIR) Post #ATDIOmbn8CkdYhZQZ6 by deprecated_ii@poa.st
2023-03-02T18:31:07.107857Z
18 likes, 9 repeats
this sort of thinking is common and is strong evidence that while "software engineering" may be a real thing, the software industry is largely devoid of engineers
(DIR) Post #ATDId3fGVwNiVl0CLg by Inginsub@clubcyberia.co
2023-03-02T18:33:35.331134Z
9 likes, 0 repeats
@deprecated_ii >performance is not a priority>gamedev
(DIR) Post #ATDIsvXqTF7bxh6AOe by apropos@freespeechextremist.com
2023-03-02T18:36:34.471903Z
12 likes, 1 repeats
@deprecated_ii there was a brief period where hardware was getting so much better, so fast, that someone throwing some shit together with Perl could beat someone putting in heroic efforts into C because the Perl app would get customers and then be able to buy next year's hardware with the income, completely blowing away any performance concerns.That period broke programmers' minds. It was a day of intense heat in the middle of winter, that has nobody caring about winter conditions anymore because all they can remember is the one hot day.
(DIR) Post #ATDIwJe9JkwVa3fANU by Binkle@sleepy.cafe
2023-03-02T18:37:09.627617Z
5 likes, 0 repeats
@Inginsub @deprecated_ii
(DIR) Post #ATDIz8Khq0N7MQubrs by apropos@freespeechextremist.com
2023-03-02T18:37:41.768583Z
7 likes, 0 repeats
@deprecated_ii and no this guy is not at all tying his remarks about performance to hardware getting better, but that one period of hardware getting better is *why* he has this position now.
(DIR) Post #ATDJ8GZUZoVTWCBPUm by PunishedD@poa.st
2023-03-02T18:39:20.565065Z
6 likes, 3 repeats
@deprecated_ii Devil's advocate: he's right that a lot of software writing doesn't care about performance... because the managers and clients don't care about performance. I haven't read the referenced post, but I bet somewhere in there is the implication that software engineers should care about it anyway and force it to be part of a dev project, no matter how much the client squeals. That works sometimes, but not always.Coincidentally, last night I poasted about and re-read The Story of Mel. This was posted to Usenet as an old legend back in 1983. It struck me that we are already 2 human generations, probably 3, away from the era of actual hardcore engineering.catb.org/jargon/html/story-of-mel.html
(DIR) Post #ATDJ9AiHqIoddgjDyi by Moon@shitposter.club
2023-03-02T18:39:28.444018Z
4 likes, 1 repeats
@deprecated_ii it's a business and delivery is the most important thing but uh software engineers fail that all the time too lol
(DIR) Post #ATDJUNu7zgdadDNkVE by Ivan_Ivanovich@poa.st
2023-03-02T18:43:20.457121Z
6 likes, 1 repeats
@deprecated_ii >in many domains, run-time performance is not a prioritythat's a self fulfilling prophecy and the reason everything runs like shit nowThis pajeet is invoking the wisdom of some objective and "expert" third party that doesn't exist which decides whether performance a priority or not
(DIR) Post #ATDJWQkGP1Cy1OkAG8 by MeBigbrain@poa.st
2023-03-02T18:43:42.538050Z
2 likes, 0 repeats
@deprecated_ii >performance is not a priority...why?
(DIR) Post #ATDJg0pnaoa2YZtyM4 by Ivan_Ivanovich@poa.st
2023-03-02T18:45:26.493669Z
7 likes, 1 repeats
@deprecated_ii yeah bro I'm sure your shit code that selects for how fast you can release that product (which will run like shit) is all about "readability and maintainability", i.e. patching all the bugs that should have been fixed before your rushed release
(DIR) Post #ATDJwoPF3QTWvxUTcu by MissionFailure@poa.st
2023-03-02T18:44:36.874911Z
3 likes, 0 repeats
@deprecated_ii Just like the generational theory, entertainment and technology follow.
(DIR) Post #ATDKMivM3iQ5ScrneK by deprecated_ii@poa.st
2023-03-02T18:53:09.252069Z
9 likes, 3 repeats
@MeBigbrain because reasonably performant code requires engineers who know what they're doing, and that does not describe the vast majority of people working in software
(DIR) Post #ATDKUGznmO3R9wXxYW by RealAkoSuminoe@poa.st
2023-03-02T18:53:18.578248Z
3 likes, 0 repeats
@deprecated_ii Here's the article in question. He's not wrong about some of these abstraction techniques causing worse performance, but at the same time the issues he brings up are usually not the reason that code is slow in practice.computerenhance.com/p/clean-code-horrible-performance
(DIR) Post #ATDKZIVRaj9jI8k7YO by deprecated_ii@poa.st
2023-03-02T18:55:25.651505Z
0 likes, 0 repeats
@RealAkoSuminoe pretty sure he knows what he's talking about
(DIR) Post #ATDKyxLWszZNC7LAUC by PunishedD@poa.st
2023-03-02T19:00:04.051828Z
1 likes, 0 repeats
@deprecated_ii @MeBigbrain Put another way, when you paid for The Average Software Engineer, you used to get someone who wrote performant code as a matter of habit. Or, he could recognize where the problems would be, and either raise it as a risk or plan to address it in an optimization patch later.Now, TASE doesn't even think about it. He's too busy trying to piece libraries together and configuring his environment before the sprint is over.
(DIR) Post #ATDLa5MW2uAOlnVSca by cyberdystopia@minidisc.tokyo
2023-03-02T19:06:36.929Z
2 likes, 0 repeats
@deprecated_ii@poa.st i remember when he tore the microsoft windows terminal devs a new asshole https://www.youtube.com/watch?v=99dKzubvpKE
(DIR) Post #ATDMb2Hr44X5TaLLjE by pro@mu.zaitcev.nu
2023-03-02T19:18:05.350200Z
3 likes, 0 repeats
@deprecated_ii @MeBigbrain I suspect he's just throwing grand pronouncements for the style. Of course a minimum level of performance is required, but it's also the case that ultimate performance may not be necessary (better programmers than that dolt made such statements on various occasions).My personal problem with performance of modern systems is how unpredictable and fragile it is. There is essentially no way to make choices based on expected performance when programming anymore, except in the broadest terms. So we give up on that. Just program it in any random way, later measure, optimize. Result is, any change in circumstances can tank performance of an existing system.
(DIR) Post #ATDMizdWnpnjZZH7SK by CathodeRayGun@poa.st
2023-03-02T19:07:02.621129Z
0 likes, 0 repeats
@MeBigbrain @deprecated_ii Example: Code for multidisciplinary applications does not need to be fast. It doesn't matter how fast your code is when you're waiting on technicians flipping switches or motors rotating or power electronics reaching thermal stability.If performance was always a priority, then everyone would write machine code. Interpreted languages would not exist due to lack of userbase. More often than not, deadlines are more important than performance, and sometimes more important than maintainability. Unfortunately.
(DIR) Post #ATDMj0F6Y7rJS6r81g by deprecated_ii@poa.st
2023-03-02T19:19:34.865963Z
1 likes, 0 repeats
@CathodeRayGun @MeBigbrain In practice what's happening in those scenarios today is people are injecting a bunch of extra hardware complexity into the system to support the abstraction they need to get away from writing C or, heaven forbid, assembly. That will not end well.And "performance is not a priority" is misleading. The real position of most programmers is performance *doesn't matter*. But of course it does, even if the constraint is easy to meet.
(DIR) Post #ATDMptOmOSJWFlfNtA by PoalackJokes88@poa.st
2023-03-02T19:20:50.187860Z
2 likes, 0 repeats
@Ivan_Ivanovich @deprecated_ii The original Mario was only 32 kbs, smaller than the profile pictures of everyone on this site. A youtube video of the original Mario music is about 300x bigger than the entire game. We're not as limited by hardware as we were back then but the software bloat has gotten so bad it's erased almost all gains made by hardware in the last 10 years.
(DIR) Post #ATDNj9S6PMdnyBZPiS by deprecated_ii@poa.st
2023-03-02T19:30:49.311078Z
6 likes, 1 repeats
@PunishedD >...clients don't care about performance.mm well I have a problem with this line of thinking, in that it's unethical to sell someone shit just because they don't know you can make not-shit too. I've guided more than one client toward a better solution than what they originally asked for and they were not upset about it. clients don't always know what they don't knowif all a company can make is shit, well, that's a different problem
(DIR) Post #ATDNyfIoGwoN5wcZn6 by BuncyTheFrog@poa.st
2023-03-02T18:59:03.950796Z
1 likes, 0 repeats
@deprecated_ii @MeBigbrain My former job was working on an event processing engine - we had stuff written in assembly because performance was so important. Before my time, so I don't know if it was strictly necessary, but: it was fast, and not easy to follow.
(DIR) Post #ATDPWPIiN7npGcEJlY by PunishedD@poa.st
2023-03-02T19:50:56.108725Z
2 likes, 0 repeats
@deprecated_ii Yeah a good company/team should guide their clients, but ultimately it's their call. When I say "don't care", I mean "don't care enough to pay for extra dev time to make it perform better".I had a case where someone requested a system that handles 150 simultaneous users, within a certain infrastructure. We give them something that can handle 300 but warn that there's some known bottlenecks that will make it run slow at the upper limit. A year later they come complaining that it slows down around 250 users, and we respond with the professional version of "WELL DUH".We can't make them choose a better infrastructure, force the DB team to upgrade their garbage, or get around their internal network issues. We tried getting the client to let us take an extra 4 weeks to try some optimizations, but they wouldn't budge on their already aggressive deadline.So we didn't even try to optimize it. We knew it was inefficient, but we also knew it was tolerable within spec and even above spec. Successful delivery, inefficient code. That doesn't make it good, but ethically I'd call it clean because the client made the call. I disagree with the soyboy that cases like this exist everywhere, or are even the majority. They also don't eliminate the need for performant code, or the need for engineers to learn how to code efficiently by habit. But they DO reliably pop up in certain sectors. If this guy only saw these kinds of projects, he might think it applies everywhere. But an IT nerd opining on an entire industry should be smart enough to recognize that fallacy.
(DIR) Post #ATDRSW7BIOSD9N6lBw by RealAkoSuminoe@poa.st
2023-03-02T19:39:28.332539Z
3 likes, 0 repeats
@PoalackJokes88 @Ivan_Ivanovich @deprecated_ii Well... profile pictures and a video being larger than code is not new. Assets are generally a lot larger than the code itself. That's more about increases in graphical fidelity than software bloat.Also, code size will naturally be larger today than it was in the past because register size has increased. The NES would have 8-bit pointers, and modern computers have 64-bit pointers, so binaries are naturally going to get a lot larger.Not saying that software bloat isn't an issue, but...
(DIR) Post #ATEVsN8nsPhfp2PXiS by birdulon@shpposter.club
2023-03-03T08:36:43.639635Z
1 likes, 0 repeats
@deprecated_ii Engineering is all about compromises and Casey Muratori is the sort of programmer that will happily sacrifice correctness and maintainability on the altar of performance. Some domains would prefer a trivial job run all day than for it to crash fast or produce invalid results.
(DIR) Post #ATEWkS4TlJXQkbAvzM by birdulon@shpposter.club
2023-03-03T08:46:29.326985Z
1 likes, 0 repeats
@MeBigbrain @deprecated_ii You get what the customer pays for. Often that involves constantly changing requirements that invalidate many inflexible micro-optimisations. Computer-hours are often substantially cheaper than man-hours so optimisation for the sake of optimisation doesn't really pay the bills for a lot of business2business software.
(DIR) Post #ATEg7P8nipr3uVDTuq by MeBigbrain@poa.st
2023-03-03T10:31:36.472494Z
2 likes, 0 repeats
@birdulon @deprecated_ii "You get what you pay for" doesn't apply to monopolies. You get what the overlords feel like giving you and pay what they feel like charging you.
(DIR) Post #ATEjSccR6e1Tegt3Dc by birdulon@shpposter.club
2023-03-03T11:08:54.395057Z
0 likes, 0 repeats
@MeBigbrain @deprecated_ii Sure, and monopolies have no reason to spend man hours on optimisation with no hope of return either.
(DIR) Post #ATEpVsXwwYgTeDizM8 by Foil_Hatter@chudbuds.lol
2023-03-03T10:51:32.313543Z
1 likes, 0 repeats
Unfortunately I can say the same about HW engineering - lots of people do it, not many are engineers more than on paper. I'd argue that it's like that in any profession. If bell curve applies to intelligence distribution in general, why would specific professions be different? What scares me though, is that if it's true, same applies to doctors. A thing to remember at your next visit, huh?
(DIR) Post #ATEpocUwpqitNPG00G by deprecated_ii@poa.st
2023-03-03T12:20:15.821978Z
0 likes, 0 repeats
@birdulon he really isn't
(DIR) Post #ATEpuEnlqoYTcYjwem by birdulon@shpposter.club
2023-03-03T12:21:09.704265Z
0 likes, 0 repeats
@deprecated_ii He literally did on the windows terminal rewrite until enough people badgered him on it
(DIR) Post #ATEqJOe8NzltOQEHDc by deprecated_ii@poa.st
2023-03-03T12:25:49.513032Z
1 likes, 0 repeats
@birdulon you mean people nitpicked example code, which is the same thing that happened with the clean code examplevery rarely do casey's critics ever bother understanding his whole approach to programming, instead preferring to look at what are essentially case studies in a vacuum. I'm pretty familiar with what he does and I'll say it's superior to almost everything I've seen from the people shitting on him for taking heterodox opinions. my project maintainability has improved significantly by adopting some of his core techniques
(DIR) Post #ATEr8yQZ71758QTb3w by deprecated_ii@poa.st
2023-03-03T12:35:08.752925Z
5 likes, 0 repeats
@birdulon @MeBigbrain Very often the horrific performance comes not from a lack of "micro-optimizations" but from companies simply choosing the wrong tools. You can't optimize your way out of an interpreted language being slow as shit. You can't make Electron not be terrible. Choosing better tools can easily mean a 50x performance improvement with very little extra effort.If performance just isn't a consideration *at all*, then these minimal effort wins are left on the table.
(DIR) Post #ATH2WpuPvHr3TdwMMK by Agartha_Noble@amala.schwartzwelt.xyz
2023-03-04T05:12:59.837090Z
1 likes, 0 repeats
@deprecated_ii This is from that dude who runs Molly Rocket right? I like him, one of the few guys who has his head on straight.
(DIR) Post #ATH2Xft8fHjIgjk4aO by Agartha_Noble@amala.schwartzwelt.xyz
2023-03-04T05:12:05.863147Z
1 likes, 0 repeats
@deprecated_ii @MeBigbrain @birdulon Fuck, there's a dozen languages that AREN'T PYTHON that can do this, even some new ones that emerged within the last 20 years, but Noooo, We HAVE to keep using fucking Java, and Javascript, and fucking PYTHON in deployment. Python is for TINKERING. If you are DEPLOYING PYTHON, you are doing SOMETHING WRONG.