What is a sufficient length for a stream? What is too short?

I'm never catching anything live so agree with what others have said, I'll just watch the stream at my own pace. Only thing I'd mention is the FAF Live channel has a lot of quite short streams as there's typically a restart or two due to technical issues. I understand this is because the stream is used by many different people so nobody is proficient, but working out those issue before starting would give a more polished feel.
Perhaps just a checklist provided by one of the experienced streamers to go through before beginning could help? (as Jip has suggested)

When it comes to caster style, do you prefer a fun and entertaining personality or a deeply-analytical one?

Considering FAF Live typically has multiple casters at once I think the most important thing is for them to have an entertaining dynamic. CaptainKlutz and Jip did a good job of bouncing off each other for the Settling Scores cast. A pattern I've seen in other RTS casts is to have a "colour" caster paired with an analytical one, since that leads fairly naturally to a fun dynamic (as FTX suggested).
Another note on polish would be for the casters to take opportunity of 'empty' moments to introduce themselves to the chat, since people (or I at any rate) usually will know names but not voices.

What kind of content do you want to see on FAF Live? Strictly high-level play, a mix of high-level and average-joe events, anything and everything possible (the more the merrier)

I personally prefer high-level, but think a mix would be best for the audience more broadly. Another consideration is that the players, casters and tournament directors are mostly drawn from higher level players, as they're the ones volunteering their time what they want to do is probably more important than what the audience to see. And of course as most FAF Live coverage has mostly been high level play to this point, this is also what the existing audience has self-selected for.
The FFA tournament a few months back was interesting to me, as an example of non high level 1v1 gameplay.

So it seems that the main argument in favour of changing is increased readability, with the benefit being mostly for newer\less attentive players as the timing and behaviour of the unit will quickly distinguish whether it's a Hunter or a Mantis.

The main con seems to be taking away factional asymmetry, a lack of which is a pretty common talking point for SupCom. In my opinion the more asymmetry the better, so long as it's balanced.

I think this "behaviour reading" is a skill that is satisfying to learn, and removing the need for it would lower an early milestone for figuring out the game (as a pretty new player myself). While in general I dislike "ivory tower design" where things are difficult to learn just to reward players who learn the game, I think learning to read the game state zoomed out is an integral and distinctive part of the game, and learning to distinguish Hunters from Mantis is one of the first ways a player can utilise that skill.

Disregarding the "cybran disinformation design" discussion as I don't have any new points to raise (and as it's an aesthetic instead of a balance question), I think the main question is instead: is confusion whether it's Mantis or Hunters disproportionally powerful? I don't think so for the aforementioned "attentive players will know" and also because labs aren't that relevant beyond the early game. A mixed force of Mantis and Hunters is a bit of a silly hypothetical in my opinion, an equivalent mass force of pure tank will beat it anyway.

What is a sufficient length for a stream? What is too short?

I'm never catching anything live so agree with what others have said, I'll just watch the stream at my own pace. Only thing I'd mention is the FAF Live channel has a lot of quite short streams as there's typically a restart or two due to technical issues. I understand this is because the stream is used by many different people so nobody is proficient, but working out those issue before starting would give a more polished feel.
Perhaps just a checklist provided by one of the experienced streamers to go through before beginning could help? (as Jip has suggested)

When it comes to caster style, do you prefer a fun and entertaining personality or a deeply-analytical one?

Considering FAF Live typically has multiple casters at once I think the most important thing is for them to have an entertaining dynamic. CaptainKlutz and Jip did a good job of bouncing off each other for the Settling Scores cast. A pattern I've seen in other RTS casts is to have a "colour" caster paired with an analytical one, since that leads fairly naturally to a fun dynamic (as FTX suggested).
Another note on polish would be for the casters to take opportunity of 'empty' moments to introduce themselves to the chat, since people (or I at any rate) usually will know names but not voices.

What kind of content do you want to see on FAF Live? Strictly high-level play, a mix of high-level and average-joe events, anything and everything possible (the more the merrier)

I personally prefer high-level, but think a mix would be best for the audience more broadly. Another consideration is that the players, casters and tournament directors are mostly drawn from higher level players, as they're the ones volunteering their time what they want to do is probably more important than what the audience to see. And of course as most FAF Live coverage has mostly been high level play to this point, this is also what the existing audience has self-selected for.
The FFA tournament a few months back was interesting to me, as an example of non high level 1v1 gameplay.

Glad to hear it.

"java -version", it stills says I'm running java 16.

That would be because you still have whatever other java you had installed in your path environment variable (echo $PATH if you're curious) but the openjdk15 archive you extracted isn't. Openjdk15 won't interact with the rest of your system at all unless you explicitly tell something to, like you have with the FAF client. Adding the java install path to the path environment variable is the sort of thing packages usually do but since you 'manually' installed openjdk15 here it didn't happen.

Sorry if I'm over explaining, just trying to "teach a man to fish" as it were.

I moved away from Ubuntu pretty quickly because the PPA system never worked very well for me.

You could just untar this somewhere and point the client to it. https://jdk.java.net/15/

Not sure about the ubuntu package but for Arch\Manjaro you need to edit the downlords-faf-client script to correctly point to your java 15 install.

Change INSTALL4J_JAVA_HOME to wherever your java is.

So it seems that the main argument in favour of changing is increased readability, with the benefit being mostly for newer\less attentive players as the timing and behaviour of the unit will quickly distinguish whether it's a Hunter or a Mantis.

The main con seems to be taking away factional asymmetry, a lack of which is a pretty common talking point for SupCom. In my opinion the more asymmetry the better, so long as it's balanced.

I think this "behaviour reading" is a skill that is satisfying to learn, and removing the need for it would lower an early milestone for figuring out the game (as a pretty new player myself). While in general I dislike "ivory tower design" where things are difficult to learn just to reward players who learn the game, I think learning to read the game state zoomed out is an integral and distinctive part of the game, and learning to distinguish Hunters from Mantis is one of the first ways a player can utilise that skill.

Disregarding the "cybran disinformation design" discussion as I don't have any new points to raise (and as it's an aesthetic instead of a balance question), I think the main question is instead: is confusion whether it's Mantis or Hunters disproportionally powerful? I don't think so for the aforementioned "attentive players will know" and also because labs aren't that relevant beyond the early game. A mixed force of Mantis and Hunters is a bit of a silly hypothetical in my opinion, an equivalent mass force of pure tank will beat it anyway.

I had to change line 8 to match where my SupCom was installed (appears to be a typo there anyway, "steampps"). Otherwise works well, I was getting atrocious performance in the score screen on Proton 5.0

@tatsu said in Linux Support:

works brilliant

it's FA in offline mode.

Don't have any use for that but if I start wanting to play Ai skirmishes that I can save and load, why not

In Downlord's FAF client change in Settings -> Forged Alliance Forever -> Command Line Format for Executable to

/path/to/run_faf.sh "%s"

To use it in FAF.