Carrying the Laptop Open: Software Engineers Before vs After Agents

Description

Two-panel photo meme captioned 'software engineers before vs after agents'. Left panel: a hand carries a thin dark laptop fully closed, gripped flat like a folder, over a grass background. Right panel: the same laptop is now carried half-open in a V shape, screen and keyboard visible, dangled casually from one hand like a paperback. The joke is that since AI coding agents run autonomously, engineers no longer close their laptops - they carry them open everywhere so the agent can keep grinding through tasks unattended, with the human reduced to occasionally glancing at progress

Common Tags

AIAssistants AITools DeveloperProductivity Automation AI DeveloperLifestyle DeveloperHumor MemeFormats Hardware

Comments

115

Anonymous ★ Top Pick We finally achieved continuous delivery - the laptop lid just isn't allowed to participate in it anymore

Anonymous ★ Top Pick

We finally achieved continuous delivery - the laptop lid just isn't allowed to participate in it anymore
Anonymous

Nothing says autonomous agent like a human acting as the laptop's keep-awake daemon.
@Vlasoov 3mo

DevBox🤔
@deimossos 3mo

One line change in logind.conf to not sleep when closing the lid...
1. @death_by_oom 3mo
  
  That's a MacBook 🤓☝️. Also doing this generally on any laptop is a bad idea
  
  @deimossos 3mo
  
  Don't care, this laptop will suffocate in a wardrobe as a homelab
  
  @f0cu53d 3mo
  
  Why
  
  @DerKnerd 3mo
  
  Usually, the thermals need an open lid
  
  @f0cu53d 3mo
  
  Ah, that. Modern laptops are cold especially macbooks
  
  @DerKnerd 3mo
  
  depends, many fans do pull air from under the lid though
2. @RiedleroD 3mo
  
  alternatively, take the magnets out :)
  
  @death_by_oom 3mo
  
  MacBooks don't use magnets anymore, there is an angle sensor in the hinge afaik
  
  @RiedleroD 3mo
  
  because we gotta reinvent the wheel every time it works for too long
  
  @death_by_oom 3mo
  
  I think it's used for keyboard under light timings, open closed animations and so on. Not sure why they did that exactly
  
  @RiedleroD 3mo
  
  because apple is a vanity brand that does useless stuff to seem more special than they are
  
  @death_by_oom 3mo
  
  M chips are a miracle of modern engineering and it's objectively the best modern portable device chip on the market, I think they get a pass on that
  
  @RiedleroD 3mo
  
  yeah wow they accidentally made something good. big deal. doesn't change the fact that most of what I despise about modern hardware was normalized by apple in their endless pursuit of looks over functionality
  
  @death_by_oom 3mo
  
  MacBook keyboards are better then most laptops and the other copy them, they set the trend for actually good trackpads, is still can't find a full machined aluminium body for a sane price anywhere else
  
  @RiedleroD 3mo
  
  I've used a macbook keyboard before and I strongly disagree. lenovo has much better keyboards on average. and let's not forget the QA issues, with keyboards breaking all the time in certain laptop models
  
  @death_by_oom 3mo
  
  Which MacBook? They changed them since the last gen intel macbooks
  
  @RiedleroD 3mo
  
  various. one of my friends has a 2012 model, and another one bought one in 2024. in school I also occasionally typed on other peoples' macbooks. all of them feel exceptionally cheap and shitty
  
  @RiedleroD 3mo
  
  I'll say the new model does feel nicer than the 2012 one, I guess
  
  @death_by_oom 3mo
  
  They 2012 has a shitty keyboard, agreed. I have one of those, I use it as a homelab. But it's been 14 years, the new keyboards are a heaven to type on
  
  @death_by_oom 3mo
  
  That may be personal preferences, but I hate travel on laptop keyboards, imo they should have as little as possible travel
  
  @RiedleroD 3mo
  
  oh I agree. I'm not a huge fan of desktop keyboards for that exact reason. but apple does it badly imo. dunno how to quantify it really, it just feels more similar to a shitty dell than a thinkpad I actually enjoy typing on
  
  @RiedleroD 3mo
  
  as I said, I've felt those ones too, and I still prefer a good thinkpad keyboard over that. even my badly built previous laptop was better
  
  @death_by_oom 3mo
  
  I really don't get it, I typed on a lot of thinkpads, acers, hps, and so on, they are all so flimsy, the keyboard caves in when you type, and the feeling of it is obnoxious
  
  @RiedleroD 3mo
  
  ok look, you can't compare apple to literal shovelware and say it's good. acer and hp make terrible laptops. thinkpads have a lot of variation for some reason; in general the older ones are better, and among the newer ones the midrange feels best. high-end (1000€+) feels like apple, and low-end (<300€) feels like dell.
  
  @death_by_oom 3mo
  
  I don't remember which laptop it was exactly, but it was some windows-running high tier Ryzen laptop for around the same money as a MacBook air, and I hated every second of typing on it
  
  @RiedleroD 3mo
  
  a thinkpad? yeah the high-end ones have kind of bad keyboards. I think they're trying to emulate apple and uh, failing.
  
  @RiedleroD 3mo
  
  this is generally true for a lot of the non-gaming industry, not just lenovo
  
  @cafeed28 3mo
  
  who makes good laptops then
  
  @RiedleroD 3mo
  
  really good ones? nobody. lenovo still makes some ok ones. some other brands probably too but I haven't looked in a while.
  
  @ZmEYkA_3310 3mo
  
  Lenovo chinese laptops
  
  @cafeed28 3mo
  
  xiaoxin?
  
  @ZmEYkA_3310 3mo
  
  Yuh
  
  @theVakhovskeIsTaken 3mo
  
  has a subpar keyboard compared to LOQ Gen9/Legion/Thinkpad X/X Carbon, it's not even an actual Lenovo product but an OEM rendition
  
  @ZmEYkA_3310 3mo
  
  Idk i carry a 60% with me around
  
  @Broken_Cloud_1 3mo
  
  I got one and it was fine
  
  @RiedleroD 3mo
  
  also I genuinely don't like aluminium bodies. it's a laptop; I don't wanna burn my thighs if I put it on my lap.
  
  @death_by_oom 3mo
  
  And I don't like to be able to bend my laptop with my hands
  
  @Art3m_1502 3mo
  
  How do you open it then
  
  @RiedleroD 3mo
  
  nor do I. good plastic doesn't do that. bad aluminium does (my previous laptop was all-aluminum and bent in all silly directions…)
  
  @Ajay_Jammu 3mo
  
  Which laptop you use?
  
  @RiedleroD 3mo
  
  and the M chip wasn't made to be fast. it was made to be exciting and new. it's only good because it's a high-budget ARM
  
  @death_by_oom 3mo
  
  It has the best balance of performance and battery life out of all the laptops
  
  @Algoinde 3mo
  
  no longer true once framework 13 pro drops
  
  @RiedleroD 3mo
  
  framework is also overhyped, but I won't get into that. either way, I'm glad apple (and framework?) is part of making ARM for desktops finally a thing. x86 shouldn't have become the standard for as long as it has
  
  @feedable 3mo
  
  arm is also not very good tbh imo
  
  @Art3m_1502 3mo
  
  But it's better
  
  @feedable 3mo
  
  idk, seems like xchging blob-ridden mess by a blob-ridden mess
  
  @Art3m_1502 3mo
  
  You should make changes to improve. There won't be a hardware Jesus that will make perfect hardware and TempleOS 2
  
  @feedable 3mo
  
  you can do nothing about hardware unless you are a hardware manufacturer but even then, the hardware landscape of today is a blob-ridden mess
  
  @RiedleroD 3mo
  
  I mean… mobile ARM are terrible. mostly because mobile is terrible in general though.
  
  @feedable 3mo
  
  both mobile and desktop arm is terrible because of their driver/firmware zoo, with most of it being proprietary undocumented bullshit
  
  @RiedleroD 3mo
  
  desktop arm is terrible because it's based on mobile arm afaik. also the hardware is just terrible, not just the docs. especially on the GPU side of the SoC, but that's not technically ARM, I guess.
  
  @feedable 3mo
  
  i am also talking about apple here
  
  @RiedleroD 3mo
  
  I don't know much about that side of apple, sorry. I was mainly talking about the snapdragons they made for ARM windows machines
  
  @feedable 3mo
  
  i mean, it is moistly the reason why asahi is so slow in terms of features
  
  @RiedleroD 3mo
  
  I guess?
  
  @RiedleroD 3mo
  
  nothing is perfect, but it's certainly better than x86
  
  @dev_key 3mo
  
  ARM uses a weak memory model, meaning stores can be reordered after loads and vice versa. Multithreaded (not single threaded) code that works on x86 could throw random bugs because of incompatibility with ARM.
  
  @feedable 3mo
  
  if you are compiling from source, then the software is at fault for having a bug there. If you are running an emulator, it should deal with that stuff by itself
  
  @dev_key 3mo
  
  well if a software was written for x86 CPU specs and there only "stores after reads" could be re-arranged. x86 - only 1 direction of re-arrangement is allowed. arm - 4 directions. meaning - same code works on x86 and could have problems on ARM. Note: again, only multithreaded code is in question.
  
  @feedable 3mo
  
  that wouldn't matter though for for anything that's not x86 assembly basically, no?
  
  @dev_key 3mo
  
  I'm talking abt re-arrangements by CPU (it's CPU's specs basically), not the compiler. Or maybe I don't understand your point.
  
  @feedable 3mo
  
  yes, and I'm saying that you normally shouldn't be able to write code "for x86" that will fail like that without writing x86 assembly directly
  
  @dev_key 3mo
  
  no no you totaly can. once again: for ex. I write code that is taking for consideration only re-arrangement in 1 direction. C++ have memory barriers to deal with it. this code could fail on ARM. Since ARM could re-arrange reads and writes differently.
  
  @feedable 3mo
  
  Yes, and C++ compilers can also perform reordering during translation, so such code is likely to fail for x86 too
  
  @dev_key 3mo
  
  for x86 if I use memory barrier - then compiler will consider it as well as CPU. in other words, there will be NO unwanted rearrangements.
  
  @feedable 3mo
  
  Yes, and same for arm
  
  @feedable 3mo
  
  arm does have barrier insns too
  
  @dev_key 3mo
  
  yes. but my point is: I will need to EDIT that code that works on x86. since "rules for ARM are different". I will be forced to ADD more barriers.
  
  @feedable 3mo
  
  No, those barriers will have to be present for x86 too. These barriers also prevent reordering during translation. Without them, the insn may get reordered anyway. The ex86-specific guarantee is not exposed to the lanmguage, onlty to the backend. In other words, YOU have to place the barriers for both. The COMPILER when emitting assembly doesn't
  
  @feedable 3mo
  
  C++ doesn't really care about what the underlying CPU memory model is, it only cares about its notion of happens-before that it itself defines
  
  @feedable 3mo
  
  Atomic reordering is an issue that is entirely contained to the backend, it does not affect the language
  
  @feedable 3mo
  
  In total, this is basically false as written: Rules in both cases are rules for the C++ abstract machine, which comp[ilers for arm and x86 both implement. But C++ doesn't care HOW the implementations do it, what matters is that both impls conform, and that the code conforms to the C++ rules, not arm or x86 rules
  
  @RiedleroD 3mo
  
  …ok. and I'm sure x86 code doesn't translate 1:1 to powerPC either. big surprise, architectural changes mean you need to change your software too
  
  @Algoinde 3mo
  
  it may have been considered overhyped 2 years ago, but now 5 years later when you can literally update the first model they released to the new aluminum macbook-like chassis with better battery, *or* drop in the new panther lake into your own 5 year old chassis?
  
  @RiedleroD 3mo
  
  it's overpriced to all hell
  
  @MarioRossi8377391 3mo
  
  (Un)fun fact, afaik this angle sensor is paired to the machine and changing it requires recalibration from Apple
  
  @MarioRossi8377391 3mo
  
  But at least we can play it as an accordion /s https://github.com/bruiselea/MAcordion
@koyokeyo 3mo

caffeinate
1. @death_by_oom 3mo
  
  caffeinate doesn't prevent hardware sleep, only idle sleep
@keftcha 3mo

doing this because they couldn't configure their OS
@RiedleroD 3mo

bs. I gotta keep those torrents running
Егор 3mo

This conversation is a better meme than meme.
@dev_key 3mo

Barriers for other directions.
@dev_key 3mo

iirc barriers have directions I might not need certain barriers when CPU spec don't ask for them.
1. @feedable 3mo
  
  No, this is false. You still do.
  
  @feedable 3mo
  
  The compiler backend may decide not to emit an instruction for the barrier if it's already satisfied, but that's it
  
  @feedable 3mo
  
  You still need the barrier in the program translated
2. @feedable 3mo
  
  Please consult eel.is/c++draft/intro.races and observe that there is nothing that refers to synchroniztion rules being implementation-defined in any way
  
  @dev_key 3mo
  
  I was under impression from this article (also there's YT video in english for those who're interested) https://habr.com/ru/companies/jugru/articles/541362/
  
  @feedable 3mo
  
  ah the classic ub
  
  @feedable 3mo
  
  the funnier thing is that the article itself doesn't even mention target differences
  
  @feedable 3mo
  
  moreover, the article says basically the same thing i have been saying: Компьютер может переставить эти операции местами: сначала обновить locked, потом положить новый элемент в очередь. Вы можете сказать: «Это неправильная оптимизация, это сломает мой код, и вообще, у компьютера не должно быть возможности так делать!» Но эта оптимизация такая же, как и в предыдущих примерах. Процессор или компилятор могут изменить порядок выполнения связанных чтений и записей, потому что решат, что это ускорит некоторые части вашего кода. Как показано в предыдущем примере, это может полностью нарушить ваши планы, но это только из-за того, что вы не сообщили о них компьютеру.
  
  @feedable 3mo
  
  pay attention to this part
  
  @dev_key 3mo
  
  yeah... but there's also a table with comparsion of x86 and ARM. and an example where the same code executes diff on iPhone and PC.
  
  @feedable 3mo
  
  Yes, that example is incomplete: spinlock is not a thing
  
  @feedable 3mo
  
  Given that the article later mentions the implementation with memory_order_relaxed, no wonder this doesn't work: the writes to the volatile variable are not synchronized
  
  @feedable 3mo
  
  Also the example uses volatile to inhibit compiler reordering, which that version of the compiler obliges to do, but it is still undefined, on both x86 and arm. The fact that he got the right result here is a mere coincidence
  
  @dev_key 3mo
  
  will read that later, thank you
@dev_key 3mo

This convo is indeed pretty interesting!
1. @RiedleroD 3mo
  
  crazy thotbot react
dev_meme 3mo

wow what a fight over a macbook best price per performance ever for me in 2018-2019 was Xiaomi, literally macbook which is still going stronger than ever, since that I always recommend for everyone who wants a laptop and not a mac to get something chinese - still beating the whole market for usual non perf user
@H3R3T1C 3mo

Hey the M chips are good because ARM and unified RAM, but a high percent is because the software and the OS running on these machines are special and 100% focus on that hardware with a full low level control of every aspect
@H3R3T1C 3mo

Try to do that with the PC high fragment ecosystem?
@H3R3T1C 3mo

Every brand want to create a custo acpi table with custom actions, then only generales shitty dsdl code and only for windows
@H3R3T1C 3mo

Etc etc.....
@cafeed28 3mo

i feel like asus occasionally does produce not bad products
@cafeed28 3mo

but asus is
Егор 3mo

This conversation is still a better meme than meme.
dev_meme 3mo

mind to explain?
1. @deimossos 3mo
  
  Agent running in the terminal or somewhere else -> the process will stop if the lid is closed, since the laptop will go to sleep

Use J and K for navigation

Carrying the Laptop Open: Software Engineers Before vs After Agents

Description

Categories

Common Tags

Comments

More like this

Usage Reset Joy, Then the Five-Hour Cap Returns

LLM Shoggoth, Agent Harness, Your Code

Agent Fixes the Bug By Growing a Body