Re: Running SuperColliderAU

Hello Brian,

Thanks for catching that. Sure enough, the "about" pane shows the message
"Built from branch 'develop' [18c4aad36]." I do think that the
documentation probably needs some updating, because when I pulled it, I
just used

git clone --recursive https://github.com/SuperCollider/SuperCollider.git

as indicated in the documentation. It's been a while since I last
built SC from source, and that was on Linux, but I don't recall the
develop branch being the default in the past.  In any case, whether it
was just my own lack of attention or whatever, I hadn't actually
intended to pull a dev branch. But it sounds like I helped find a bug
in any case!


On Tue, Mar 30, 2021 at 12:00 PM <[email protected]> wrote:

>
> Hi David,
>
> I reviewed your earlier messages, and realized you are not running version
> 3.11.2, you are running something from the develop branch. Your earlier
> error message had this in the stack trace:
>
> Object:doesNotUnderstand
> arg this = nil
> arg selector = 'protocol'
> arg args = [*0]
> Meta_Server:remote
> arg this = <instance of Meta_Server>
> arg name = 'scau'
> arg addr = <instance of NetAddr>
> arg options = nil
> arg clientID = nil
> var result = <instance of Server>
>
> But options.protocol isn't called on the options object in 3.11.2:
> https://github.com/supercollider/supercollider/blob/Version-3.11.2/SCClassLibrary/Common/Control/Server.sc#L380-L385
>
> It is after this commit on develop from a few months ago:
> https://github.com/supercollider/supercollider/commit/c1eb86075c77293c29b9cb6020f06d7e5132f9fb
>
> If you want a stable build, and one that actually corresponds to an
> "official" release, build from a version tag. This is generally a good rule
> for compiling any git-managed project.
>
> There are a few takeaways I see:
> - the documentation could be clearer about how to make a stable build and
> what the difference is
> - this is a bug in Server.remote that should be reported and fixed
> - when asking / answering what version of SC someone is running, we should
> be very explicit and if the project was compiled on the user's machine,
> *always* use the commit hash -- there is too much room for confusion here
> otherwise. (This can be found in the IDE by opening the "about" dialog.)
> - we should adopt the "coarse versioning" scheme in Marcin's RFC, which
> would have made it clear from the start that you are not running 3.11.2,
> but in fact "3.12.0-dev":
> https://github.com/supercollider/rfcs/blob/51f246cd6e820ab10161cd25f8a6876b9a91f15b/rfcs/0016-scversion-improvements.md
>
> Regards,
> Brian
>
> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
> On Tuesday, March 30, 2021 3:15 PM, <[email protected]> wrote:
>
> Hello Brian,
>
> No, Server.defaultOptionsClass does not seem to be nil. If I just execute
> Server.defaultOptionsClass;
> it returns
> -> ServerOptions
>
> Server.all;
> returns
> -> Set[ localhost, internal ]
>
> -David
>
> On Tue, Mar 30, 2021 at 8:52 AM <[email protected]> wrote:
>
>> Is Server.defaultOptionsClass is nil for you after startup? I asked this
>> in my previous email. I don't want to assert that what I say is the real
>> problem without an answer to this.
>>
>> Regards,
>> Brian
>>
>> -------- Original Message --------
>> On Mar 30, 2021, 2:47 PM, < [email protected]> wrote:
>> Hello Christof and Brian,
>>
>> No errors on startup, and localhost server starts up normally with
>> s.boot. No extensions installed at the moment and no special scripts in
>> use. It's a fresh install, built from source. I don't even have any quarks
>> installed yet - wanted to eliminate any variables while playing with SCAU.
>>
>
>