Should we make v0.9.0 a very quick cycle that only adds support for the multi-root stuff? My understanding is that the next version of VS Code (to be released in a few days) will enable all the multi-root features for all users, so it would be nice if we would support that sooner rather than later.
We could then take a little more time for v0.10.0 and add more new features, do all the various other refactoring we have planned and so on.
That sounds good to me, v0.10.0 will probably require julia0.7/1.0 support so could take a bit of time...
Couldn鈥檛 we do things like the extra symbol server process on Julia 0.6? And there seem to be lots of other ideas we also could implement without going 0.7, right? What did you have in mind that would require 0.7?
I just meant that v0.7/1.0 will probably be out by the end of the year so we should support it?
Ah, ok! I love the core julia devs, but I'm not giving any weight anymore to their timelines :) I think we are best advised to just ignore julia 0.7/1.0 until they have a formal feature complete beta or something like that. At that point we almost certainly will still have a couple of months until a final release to work on compat.
Also, in my own view, I'd be very surprised if 0.7 got released anytime before the next juliacon.
@ZacLN Do you have a sense how difficult the multi-root stuff in the LS is? I think we should essentially have a v0.9.0 that only adds stuff that deals with multi-root and nothing else. Hopefully that way we can have a much shorter beta/rc period because there might be less new features that can introduce new bugs. And then v0.10.0 could add other new features again.
I need:
workspace/... handlers to do a startswith check to determine which workspace to act onAlright, all done though will need some testing
Should we cut a beta so that folks can start to test the multi-root story? We still might need to add some stuff about settings, but it might be good to have some test reports on the pure multi-root story already.
I am seeing a lot of crashes of LS on my system. But I don't see the logs, so I don't really know how to debug that stuff...
Yep lets do it. For debugging:
function Base.run(server::LanguageServerInstance)
io = open("~/jls.log", "a")
redirect_stderr(io)
while true
message = read_transport_layer(server.pipe_in, server.debug_mode)
request = parse(JSONRPC.Request, message)
server.isrunning && serverbusy(server)
process(request, server)
server.isrunning && serverready(server)
end
end
I'll add this code as the default for the debug mode for now!
maybe add a `rm("~/jls.log") at the start otherwise the file will get huge!
I was going to create a new timestamped one at every startup?
Yep, thats better
I just never get anything in these files... I think this might not work on Windows.
maybe use "w" rather than "a"? Either way, lets push a beta
Lets do this?
Yep, I'm game. I think we still need to figure out the setting issue with multi root, and then fix 1-2 bugs? I'm looking at the weave bug right now.
I鈥檓 not quite sure what needs to be done for settings as we hardly use them?
Ok, two approval pending PRs and then we could be done.
Big question is whether we also want to include https://github.com/JuliaEditorSupport/julia-vscode/pull/429? I'm pretty much done, I think... And it would just be so cool to have that out in the wild...
I say lets stick with what we have, there's a couple of other PRs waiting so we could get an rc for 0.10.0 out fairly soon?
Alright. I'll do a 0.9.0-rc build now.
@ZacLN Should I release it to the marketplace? We had a one day RC... If we find a problem we can still do a bugfix release? But releasing would allow us to move forward with all the other stuff.
Yup, pull the trigger
And I release a v0.9.1 a minute later that has an updated CHANGELOG :) I always forget that...
Most helpful comment
Ah, ok! I love the core julia devs, but I'm not giving any weight anymore to their timelines :) I think we are best advised to just ignore julia 0.7/1.0 until they have a formal feature complete beta or something like that. At that point we almost certainly will still have a couple of months until a final release to work on compat.
Also, in my own view, I'd be very surprised if 0.7 got released anytime before the next juliacon.