Ubuntu Is Looking At Offering Better WiFi Support By Using Intel's IWD

[CommunityMember]

starshipeleven i have a router with same name for the network for both 5gh and 2ghz. with IWD i was unable to connect to a specific network (2/5 ghz). so it always connected to 2ghz due to strong signal i think.

Many access points ((nearly?) all enterprise ones, and some of the pro-sumer models) provide a hint (band steering) about preferring 5GHz over 2.4GHz, which addresses many use cases (and not just for IWD, but for other clients).

And for IWD, there is the rank_5g_factor option which can address some other use cases.

It should be noted that IWD already gives 5GHz some preference over 2.4GHz when things are approximately equivalent, but, of course, details always matter, which is why additional control is provided (via the rank_5g_factor option).

[grigi]

I'm curious; did writing that make you feel like you had your bias confirmed or what was the point? I'm asking a serious question, because I see people doing this all the time, in many different contexts, and I really don't understand why they do it.

I think (s)he's alluding to the previous projects where Ubuntu went their own way when a viable solution already existed.
Some examples are:
* Bzr
* launchpad (was a leading OSS code hosting platform at a time, but the requirement of bzr killed it for many, and hosting your own was near impossible)
* Unity (pre-8 was pretty awesome, and actually blazed some trails)
* Upstart
* Mir
* Snap

Honestly, it's not that much, but they often do them for what appears to be the wrong reasons.
I also think their insistence in having a CLA that any contributor needs to sign killed many of their projects for them.

[starshipeleven]

starshipeleven i have a router with same name for the network for both 5gh and 2ghz. with IWD i was unable to connect to a specific network (2/5 ghz). so it always connected to 2ghz due to strong signal i think.

Did you try tweaking the BandModifier5Ghz value in the config? https://jlk.fjfi.cvut.cz/arch/manpages/man/iwd.config.5

it should tip the scales in favor of choosing 5Ghz networks

[starshipeleven]

I'm curious; did writing that make you feel like you had your bias confirmed or what was the point?

It's just a sarcastic way to state that Canonical has a long history of reinventing the wheel where they really should not have done that.

[sabian2008]

starshipeleven i have a router with same name for the network for both 5gh and 2ghz. with IWD i was unable to connect to a specific network (2/5 ghz). so it always connected to 2ghz due to strong signal i think.

For me setting the BSSID in NetworkManager was also useful to prevent it from scanning every 5 minutes to see if a better signal version of my network was found (for roaming purposes). This is the only way I found to play online games without constant lag spikes, although I aknowledge that this might be network adapter dependent.

[CommunityMember]

For me setting the BSSID in NetworkManager was also useful to prevent it from scanning every 5 minutes to see if a better signal version of my network was found

Both IWD and wpa_supplicant provide options to disable the background scanning.

[sabian2008]

Both IWD and wpa_supplicant provide options to disable the background scanning.

I couldn't make the option for wpa_supplicant work consistently and didn't even know about IWD (this was ~ 5 years ago). Not saying it couldn't be done, it was the only way I found. I don't know how general my experience could be, considering it was a pretty buggy adapter (in Linux at least).

[CommunityMember]

Honestly, it's not that much, but they often do them for what appears to be the wrong reasons.

They typically had at least reasonable justification at the time the decision the made(*), although they (like others before them) have sometimes been resistant(**) to review that decision at an appropriate stage to confirm it is still the right choice given time and changing events.


(*) Since it is hard to predict the future, sometimes the choices really are more of a guess about a path to take, or the need to get something done now (the perfect is the enemy of the good enough).

(**) The term sometimes used is stubborn.

[kreijack]

I think (s)he's alluding to the previous projects where Ubuntu went their own way when a viable solution already existed.
Some examples are:
* Bzr
[...]
Honestly, it's not that much, but they often do them for what appears to be the wrong reasons.
I also think their insistence in having a CLA that any contributor needs to sign killed many of their projects for them.

Even tough I agree that Ubuntu had the habit to develop new solution without a real needing of that, regarding bzr i remember a different story.

I contributed to BZR and Mercurial at the time, so I think to know quite well what was the state. Before GIT the only big options as (D)CVS were:
- cvs
- subversion
- bitkeeper (the first DCVS)

Where bitkeeper was a big step ahead the others. However the frictions between the bitkeeper license constraints and the linux community started several projects like:
- mercurial (written in python + some parts in C for performance purpose)
- bazaar-ng aka bzr (written in python, but it started from the baazar/baz project written in C)
- git (written in C + some bash scripting)

Now GIT is definitely the winner, however in the beginning had few problems: the GIT user interface was very ugly and not very usable so some wrapper were available/needed; these wrappers provided a more user friendly interfaces (someone remember cogito /CG ?). In fact the the git-smc site[*] reports

[...]But because Git was initially a toolkit for a version control system rather than a full user-friendly VCS, it has a number of subcommands that do low-level work and were designed to be chained together UNIX-style or called from scripts.[...]

What I mean, is that at the time (2005), GIT was far to be what is today; so there were a number of reason to develop alternative tools. And I think that (at the time) canonical was right starting to develop BZR. To me BZR was a lot more user friendly than GIT. I think that GIT now is far superior than the others DCVS system, but at the time nobody could know that.


-[*] https://git-scm.com/book/en/v2/Git-I...-and-Porcelain

[Danny3]

Finally!
I've been testing it last year and it worked great, until I ran into some problems with reconnecting and on next reinstall I should wait for proper integration, which I hope it will happen now.
I hope also that Intel continues to improve it and add more features to it.