IWD 1.0 Released As Intel's Wireless Daemon For Linux Systems

[loganj]

too many automations (iwd) is not always a good thing.

[pininety]

too many automations (iwd) is not always a good thing.

The idea of iwd is actually quite good, 1.0 actually seems to now work after the segfault thing was fixed so I am very much looking forward to never see wpa_supplicant again on my system!

[starshipeleven]

too many automations (iwd) is not always a good thing.

wpa_supplicant isn't going anywhere, some competition won't hurt.

[timofonic]

When will IWD be have all the features of wpa_supplicant? Are there something similar to mesamatrix.net but for IWD?

[mikelpr]

Most of the improvements can't be carried over to wpa_supplicant because of its design goals.
iwd is supposed to be a smart daemon acting on its own, wpa_supplicant is designed to be a dumb tool that has to be controlled by something else.
iwd is supposed to target only Linux so it does not need to care about abstracting around many different OS APIs, wpa_supplicant is designed to multi-platform.

wpa_supplicant alone is bullshit, you need for the very least a bunch of scripts to deal with wifi disconnect and reconnect automatically.

With iwd you can do without a higher-level network manager application for wifi

ah, I was not thinking of wpa_supplicant standalone but with NetworkManager. what advantages does it have over that combination? I meant wpa_supplicant coupled with NM as in most distros is fine, alone obviously isn't

[pininety]

ah, I was not thinking of wpa_supplicant standalone but with NetworkManager. what advantages does it have over that combination? I meant wpa_supplicant coupled with NM as in most distros is fine, alone obviously isn't

Well, it is more of a "NM is plastering over the biggest problems with wpa_supplicant" than a "The both work fine together". If you have a atomic device that can only do one thing at a time like a wifi card or a harddrive, you should really have one thing in total control scheduling access instead of just allowing multiple things to access it in any order they want. The previously linked video on iwd explains a lot about the problems they are trying to solve, I highly recommend watching it.

[starshipeleven]

ah, I was not thinking of wpa_supplicant standalone but with NetworkManager. what advantages does it have over that combination? I meant wpa_supplicant coupled with NM as in most distros is fine, alone obviously isn't

It's smaller and lighter than NM+ support stuff, also an independent project. It's following a more modular philosophy where each networking daemon dealing with a specific network technology is self-sufficient for its area so you have a more modular system overall, and NM becomes more an orchestrator than a do-it-all, and you aren't locked to it like you would if it was done with a plugin system or even core functionality. (ModemManager is also another of such daemons, which is dealing with mobile modems and works fine on its own)
This is important for embedded devices and custom setups where you just want to integrate wifi capability to a custom software running on an embedded device.

It's simpler than NM to work with, especially if you need to do something like mesh, multi-wifi AP, repeater (i.e. if you are not a client device).
Being smaller and less abstracted means it's also less likely to have bugs, and easier to comprehend and work on for contributors. Much less lines of code, as simple as that.

If all you need is connecting to a wifi router it's not going to change much, but that's far from the only usecase for wireless.

[polarathene]

it's just a management daemon (like wpa_supplicant is), performance comes from the wifi device drivers.

And yes it's not tied to any specific hardware as it is talking to kernel API to do its job

Is there much benefit to an end user then? I don't touch wpa_supplicant and NetworkManager pretty much handles the network stuff for me afaik(with integration in Plasma network systray widget).

I remember IWD getting praised and talked up in the past, I thought that was because of handling wifi more efficiently, but it seems it's more of a lib/API that unless you interact with/setup directly, there isn't much benefit to a general user?

[tildearrow]

It is actually "iNet Wireless Daemon", whatever this means...

I have been using it for a few months now, it overall works great, especially as a companion to networkd. Unfortunately, it logs close to nothing, so when it refuses to work you won't get a clue why. Documentation used to be trash as well, but is is getting better.

Yeah! At first I thought IWD meant Intel Wi-Fi Daemon, AKA only for Intel devices...

[pininety]

Is there much benefit to an end user then? I don't touch wpa_supplicant and NetworkManager pretty much handles the network stuff for me afaik(with integration in Plasma network systray widget).

I remember IWD getting praised and talked up in the past, I thought that was because of handling wifi more efficiently, but it seems it's more of a lib/API that unless you interact with/setup directly, there isn't much benefit to a general user?

You really should watch the posted video. iwd is supposed to handle wifi more efficiently, exactly as you said. For one, scans for networks will not interrupt your network anymore, reconnecting to a wifi network after suspend will be way faster etc. Heck, of of there problems atm. is that iwd is so fast that udev does not manage to rename your wifi device during boot up.