Return to Level1Techs.com

Fresh install, partitioning, and preventing bloat


#41

it’s not a direct correlation, no, but it usually means that you have more running software

there is bigger chance for a user experiencing a bug while using a bloated system in general


#42

most of your packages are installed as dependencies to other programs. Mostly libraries. Only a small portion of software will run in the background automatically, and I’m drawing a blank on examples that fall outside the category of server software.


#43

libraries are code, the so files are loaded and share between multiple programs so a bug in one .so will fuck multiple packages


to be clear, I do agree that more packages != more running software at a given time. But it can contribute to it.


#44

To change my mind on this you need to provide me an example of this actually happening.

There’s no way to change your mind because I can’t prove a negative, so I’m done arguing this.


#45

This is still anecdotal coz just n=2

My PC: 1.3k packs, 67 tasks

My laptop: ~600 packs, 35 tasks

basically half in both cases :wink:


also look at memory and cpu


#46

in one picture you’re using openbox, the other picture looks like you’re using kwin. not exactly a scientific comparison, also they’re two completely different machines.


#47

Plasma 5 vs i3

both i3 and kde serve the same purpose here, so I disagree. KDE is more bloated way to get a desktop environment vs i3 + some tools


#48

there are more obvious reasons that one system would be using more memory than simply # of install packages.


#49

num of pgks is not the point here I wanna make, the point is more bloat = less resources = shitter user experience

I say that number of packages AND each individual package resource usage adds up to your bloat value

you said that bloat is not real, I disagree and provide my reasoning why.


#50

My primary objection was using package count as a metric, not that bloat isn’t real.


#51

big ree


#52

As a fellow UniFi owner I can highly recommend you compartmentalize it in a docker container or a VM. That way you can both backup a working solution and update it without breaking the rest of your system. You can also shut it down when not in use, if you choose to do so.

Personally I run mine in a VM on my main server here at home, but a docker solution also works really well. While I haven’t personally tried it you can also put the UniFi controller in the cloud.

Best of luck!


#53

The idea of bloat is idiotic, but that’s not the primary point that I was trying to convey. I’ve basically made every argument that I can, without reiterating. No proof has been leveraged that the same system with less packages is less buggy, and I’m not convinced that one system with plenty of excess RAM runs better or worse than another system with plenty of excess RAM.


#54

I gave you proof, at least on my machines shit checks out.

I draw my conclusions from my experiences, this was so for all of my systems.

If you have spare 1000 different computers to test this on, msg me.


#55

Two machines running two entirely different environments is not a valid test.

A much more fair comparison would be the same machine, or at least the same hardware, running the exact same environment but a vanilla install for a control, and the other has a bunch of other packages installed and then measure performance differences between the two.


#56

so no xorg? :joy:

again: the test is about bloat to functionality


#57

I would say vanilla would be defined as the bare minimum to get a graphical session setup and run a few choice programs like firefox, maybe even some benchmarking software too.

Then we need to minimize variables and have a methodical approach.


#58

what would that bench again?


#59

expected use, like browsing, media playback, maybe some kernel compiling.

Measure performance use throughout and then compare any differences.

Might actually make a good L1 video.


#60

While that might be interesting, my interest is very much singular:

Features I want / bloat.

If there’s software X that can do Y, and software Z that can do Y but has larger size, eats more CPU cycles and has 5 more deps than X. Well fuck Y, Y is bloated, I’ll get X.

It’s quite simple. The whole pkg count discussion was a tangent that I didn’t really care about, but since I love arguing for no reason (a flaw of mine you might have noticed) it kinda went on for too long lol