[LAD] [LAU] Release: New Session Manager Version 1.3

Filipe Coelho falktx at falktx.com
Thu Jun 18 14:59:28 CEST 2020


On 06/18/20 13:10, rosea.grammostola wrote:
> You know I did a lot for NSM, and I'm probably the most likely the 
> most active user in the community. Involved from the start of the project.

Yes, I know.
Though for quite some years after the initial launch you were away from 
the scene, only recently coming back to check on status of linux audio 
and related things.
So please do not make it seem like you were helping all these years, NSM 
has been dormant for a while now until we (including you) recently 
starting pushing the discussion again.

Before you came back though, we already had set on NSM as the de-facto 
SM to use (which you do not seem to believe it seems).
Last year I gave a talk at Sonoj discussion JACK, and its future. I 
specifically mention that NSM is the winner of all SMs, that 
JACK-Session is being made deprecated, and congratulations to Jonhathan 
for that.
That has been my opinion for a long time, but I could never do much 
about it.
Once I became the JACK maintainer, it was easier to push into this 
direction.

Nils was also always very active in pushing the conversation regarding 
NSM, making code examples on how to use it, and keeping the NON wiki 
updated.
He made his new tools (Patroneo, Fluahjo and Vico) entirely based around 
NSM.

> I did a lot to improve the situation lately, with some success. I've 
> had contact with both of you last week (IRC and mail). Nils sent me 
> mail this week about NSM/Jonathan. Nothing about the fork. Besides the 
> fact that I understand that at some point a fork was inevitable, next 
> time you want to fork, just say it and make sure to do it in full 
> daylight.

We talked about the fork quite a few times in the #lad IRC channel, 
where most linux-audio developers hang out.
Again, the timing of the announcement is irrelevant. Might be night time 
for you, but it was still day in the American continent, as usual per 
timezones ;)

We had many times tried to make things work with Jonathan, you of all 
people should know.
If we spoke about a fork, we would fear yet another shit-storm would 
come, and we are tired of that at this point.

PS: There is no need for personal attacks here, please cut that out and 
be nice.


> On a technical level, I'm glad you are aiming to be fully compatible 
> with the original Non-session-manager (NSM). I'm just afraid that 
> you're underestimating the task at hand and the accomplishments being 
> made here though. Not giving the original author the credits he 
> deserves, might be a sign of it as well.
>
> From personal experience, I've still have to find someone else besides 
> the original author of NSM, who understands why NSM works as a session 
> manager. It works I think, because it's simple and has clear rules.

We do understand why NSM works, so much that we are actively pushing for it!
If we did not believe that NSM can be the de-facto SM, we would not go 
to all this trouble for it, insisting on it after so many years despite 
the toxic environment around it.

I also think the clear rules is what makes NSM work, not sure why would 
say I do not. Me, Nils, David and many others, we all do.
Other SMs have clear design issues, where NSM was brave enough to be 
quite strict on what applications are allowed to do, and that is what it 
makes it possible to succeed.


> I see a urge for new features, which are potentially harmful for the 
> success of NSM. When I didn't use Linuxaudio for years and restarted 
> it, it was quite a horrible experience. JACK standalone applications 
> do have all kinds of features, but where crashing on me constantly 
> (the nice thing about NSM, is that you've it back in one click). 
> Totally unusable to make music with. I'm personally not waiting for 
> new non-essential NSM features, which are making my setup less 
> predictable, more resources consuming and less stable.

Not exactly true about new features. We do not want to add new stuff 
into NSM, it is perfectly fine as it is.

We do want though to make the user experience better.
As requests for (in my opinion sensible) features to the GUI were 
ignored, and me and Nils not being comfortable developing with FLTK, the 
only solution was to try to create a brand-new GUI that we can maintain.
This will change nothing about NSM though, the protocol will remain 
exactly the same, with very tiny exceptions where unavoidable.
One such change is the highly-requested possibility of getting the NSM 
Session dir out of $HOME and into $XDG_CONFIG_HOME, there was no way to 
support this without adding a tiny new *GUI only* change.
And the *GUI only* is critical, the NSM protocol still remains 
unchanged, this is only used for "NSM Control Panels".

Stability issues of standalone applications are a bummer for sure, but 
not related to NSM.

Also, in case this was not obvious... We made the fork leaving the 
original NSM GUI untouched, it is the same original code.
So there is no need to use any new "less predictable, more resource 
consuming and less stable" control GUIs.


> Raysession, I've no confidence in it. I said enough about it. Better 
> spent your time in NSM support for clients.

RaySession has nothing to do with this announcement, let's not talk 
about it here.


> You recommend Argodejo as GUI on the github page. I've a very hard 
> time finding it better then the default NSM GUI. The simple view is 
> not that simple anymore if you've a lot of sessions. The advanced 
> view, is more complex then the original GUI, because it gives you so 
> much more information. Duplicate was renamed as 'save as', which might 
> cause dataloss for people who expect it to behave as 'save as' in 
> other applications. Might be personal preferences, but all these small 
> things doesn't make me very enthusiastic about NSM without the 
> original author.

Thanks for the constructive criticism.
The tool is very new, and I even haven't run it myself.
Please give us some time to get it nice and polished - it is not even 
released yet.


> A other related experience. Feature request for Radium. NSM support in 
> Radium, which is great. Author did implement accidentally server 
> client osc messages. As a consequence he decides to give Radium 
> session manager functionality as well. I think this design approach 
> will harm a reliable and predictable NSM session environment for the 
> user at the end.

Yes, I agree fully.
Not sure how he confused things so much, but it is not the task of the 
applications to be messing with NSM server-side business..
Basically, he did not implement NSM properly. :(


More information about the Linux-audio-dev mailing list