4 ms·

I really wish app developers would throttle their prototypes. They tend to run overpowered machines, but then they're oblivious to performance problems their ap

by dmos62 6y ago

I really wish app developers would throttle their prototypes. They tend to run overpowered machines, but then they're oblivious to performance problems their apps have. As a consequence, users have to keep playing catchup with developers' computers. I feel strongly about this. I can't say with a straight face that the growth in apps' resource consumption over the last ten years reflects in improvements in their feature-set. Feels a bit like how gold fish get big if you put them in big aquariums.

RedShift1 6y ago
This 100x. I've always said, developers should get a Pentium 3 500 MHz and 128 MB of RAM to test their software. And yes I am following my own advise, I use an old Dell with a Core 2 Duo T8100 and 2 GB of RAM to test my software on (oldest I could find at work).
- anthk 6y ago
  Today I'd choose a Pentium 4/Athlon with SSE2 and 512MB of RAM as the minimum.
throwaway894345 6y ago
I don’t think it’s developers’ devices that are driving the trend as much as economics. Resources get cheaper so people get faster computers and developers can release faster by forgoing optimizations all while delivering the same levels of performance. Of course, this excludes potential customers who can’t afford to update, but those are probably not the folks off of whom app devs are making their money. Of course, this doesn’t legitimize the decision to forego optimizations—I’m only noting that it’s probably not the performance of developer machines in particular.
- dmos62 6y ago
  I've considered that, but I don't have the impression that app development has gotten faster over the last decade. I bet there's people who can chime in here.
  - throwaway894345 6y ago
    If we're not spending the time optimizing, it must be going somewhere. If it's not going into faster, more frequent releases it might be going into improved quality (though I doubt that as well, from experience) or perhaps even just employing cheaper developers (and QA and management and design and etc) such that they produce software of the same performance in the same amount of time.
  - kodah 6y ago
    It certainly has. Frameworks like Rails or Django didn't exist, pre-built component libraries weren't even a dream yet, and everyone was trying to get software to run consistently on each other's machines. These days there's worlds of abstraction between you and those problems. Those abstractions are in transpilation, libraries, etc which all take more power to get over the problem of time.
    - regularfry 6y ago
      >> last decade > Frameworks like Rails or Django didn't exist You... might want to check the git histories. And a calendar.
      - kodah 6y ago
        You are right, they were around. They didn't exist with the capabilities or simplicity they do today though. Rails is also benefitting from performance improvements to Ruby, most recently.
        regularfry 6y ago
        The development velocity you could get with Rails was never constrained by the interpreter speed. That was part of the point, you could get a perfectly respectable project stood up on ruby 1.8.
    - pjmlp 6y ago
      Cough, Zope, AOLServer, ASP COM libraries, Vignette,...
EvilEy3 6y ago
It's not about throttle. It is incapable to run faster at the moment.
vendiddy 6y ago
Any guides or tools for doing this? Some devs on our teams now now have M1s. I worry they won't feel the pain of slow apps if they develop with such fast computers. Meanwhile we've got a number of customers who use computers that are 5+ years old.
- calvinmorrison 6y ago
  Buy old laptops and routinely test on them. We keep a lot of old devices around solely as test devices
- dmos62 6y ago
  I don't have experience doing this, but I'd experiment with VMs and their performance settings. It would be a pretty ergonomic solution. Rpi or some x86 single board computer might make sense too. I guess what's viable depends on the stack and the software itself.
mimixco 6y ago
Software will always outrun hardware, and ginormous software drives the entire existence of the hardware business. It's just how the system works. Are we getting benefits? I can think of lots of software we didn't have 10 years ago that we wouldn't want to give up just so we could keep old, slow boxes.
andrepd 6y ago
Oh absolutely. And internet speed as well. Have them use their shitty 5MB webpage over a spotty 500kbps on a phone with 1GB RAM. Maybe they'll see what it feels like to have the page load bit by bit over a minute with stuff jumping around all the time. Maybe they'll reconsider if they need 5 megs of js to display text and images then.
dan-robertson 6y ago
If you want to work on something like flutter, which will be optimised for the hardware of the day when it is done, and therefore the hardware of the future when it is being developed, then surely it makes sense to start with a prediction of how future hardware will be (eg gpu more important, possibly more pixels to push, more (not faster) processor cores, possibly higher framerates), and then try to approximate that with a powerful present-day computer. Though in this case it feels like problems might be due to running in a web browser and not being optimised.

by dmos62 6y ago

RedShift1 6y ago

This 100x. I've always said, developers should get a Pentium 3 500 MHz and 128 MB of RAM to test their software. And yes I am following my own advise, I use an old Dell with a Core 2 Duo T8100 and 2 GB of RAM to test my software on (oldest I could find at work).

anthk 6y ago
Today I'd choose a Pentium 4/Athlon with SSE2 and 512MB of RAM as the minimum.

throwaway894345 6y ago

I don’t think it’s developers’ devices that are driving the trend as much as economics. Resources get cheaper so people get faster computers and developers can release faster by forgoing optimizations all while delivering the same levels of performance. Of course, this excludes potential customers who can’t afford to update, but those are probably not the folks off of whom app devs are making their money. Of course, this doesn’t legitimize the decision to forego optimizations—I’m only noting that it’s probably not the performance of developer machines in particular.

dmos62 6y ago
I've considered that, but I don't have the impression that app development has gotten faster over the last decade. I bet there's people who can chime in here.
- throwaway894345 6y ago
  If we're not spending the time optimizing, it must be going somewhere. If it's not going into faster, more frequent releases it might be going into improved quality (though I doubt that as well, from experience) or perhaps even just employing cheaper developers (and QA and management and design and etc) such that they produce software of the same performance in the same amount of time.
- kodah 6y ago
  It certainly has. Frameworks like Rails or Django didn't exist, pre-built component libraries weren't even a dream yet, and everyone was trying to get software to run consistently on each other's machines. These days there's worlds of abstraction between you and those problems. Those abstractions are in transpilation, libraries, etc which all take more power to get over the problem of time.
  - regularfry 6y ago
    >> last decade > Frameworks like Rails or Django didn't exist You... might want to check the git histories. And a calendar.
    - kodah 6y ago
      You are right, they were around. They didn't exist with the capabilities or simplicity they do today though. Rails is also benefitting from performance improvements to Ruby, most recently.
      - regularfry 6y ago
        The development velocity you could get with Rails was never constrained by the interpreter speed. That was part of the point, you could get a perfectly respectable project stood up on ruby 1.8.
  - pjmlp 6y ago
    Cough, Zope, AOLServer, ASP COM libraries, Vignette,...

EvilEy3 6y ago

It's not about throttle. It is incapable to run faster at the moment.

vendiddy 6y ago

Any guides or tools for doing this? Some devs on our teams now now have M1s. I worry they won't feel the pain of slow apps if they develop with such fast computers. Meanwhile we've got a number of customers who use computers that are 5+ years old.

calvinmorrison 6y ago
Buy old laptops and routinely test on them. We keep a lot of old devices around solely as test devices
dmos62 6y ago
I don't have experience doing this, but I'd experiment with VMs and their performance settings. It would be a pretty ergonomic solution. Rpi or some x86 single board computer might make sense too. I guess what's viable depends on the stack and the software itself.

mimixco 6y ago

Software will always outrun hardware, and ginormous software drives the entire existence of the hardware business. It's just how the system works. Are we getting benefits? I can think of lots of software we didn't have 10 years ago that we wouldn't want to give up just so we could keep old, slow boxes.

andrepd 6y ago

Oh absolutely. And internet speed as well. Have them use their shitty 5MB webpage over a spotty 500kbps on a phone with 1GB RAM. Maybe they'll see what it feels like to have the page load bit by bit over a minute with stuff jumping around all the time. Maybe they'll reconsider if they need 5 megs of js to display text and images then.

dan-robertson 6y ago

If you want to work on something like flutter, which will be optimised for the hardware of the day when it is done, and therefore the hardware of the future when it is being developed, then surely it makes sense to start with a prediction of how future hardware will be (eg gpu more important, possibly more pixels to push, more (not faster) processor cores, possibly higher framerates), and then try to approximate that with a powerful present-day computer. Though in this case it feels like problems might be due to running in a web browser and not being optimised.