The difference between Internals, Language and Library is pretty clear to me:
Internals: stuff that is not visible from Julia at all.
Language: stuff that exists in the core language before loading sysimg.
Library: all the stuff we define and load when building sysimg.
Language and library affect everyone; Internals should only be visible in terms of performance and memory usage. Maybe there’s no point in distinguishing Language from Library, but there’s a lot of people who don’t care about internals at all.
Hi, the move to discourse has been brought to our attention at BioJulia as possibly replacing our Gitter’s and so on. I’m wondering you think focused groups like BioJulia, JuliaStats, JuliaDocs, JuliaGPU and so on should follow this trend and have their own discourse setups, or whether all of them should simply come here and do everything on julialang.discourse under the domain (btw can we have a Biological Science domain on here pleeeeze? ).
In my opinion I would like to see the communities coming together in one place, but it is up to the communities to make that decision. Speaking for JuliaGPU we are planning to promote and use https://discourse.julialang.org/c/domain/gpu.
Thank youuuuuu .
We’re currently discussing this over at: https://gitter.im/BioJulia/Bio.jl, and it’s been asked is there anything say Gitter can do that discourse can’t right now? I expect Github integration is probably possible…
Oh would you like me to give a bit of BioJulia blurb to put in the “About This Category” for Biological Science?
Discourse is more of a forum, long text discussions (with formatting, code-highlighting, etc…), whereas Gitter is more of a chat-room (ephemeral?). What level of Github integration are you looking for?
If you send me a PM with a blurb I can update the category description.
I like the general structure and agree that we should add an announcements channel.
I have one question: How should one deal with smaller topical groups? There is a small JuliaGeo community with a github organization maintaining a few Geoscience related packages and a low-traffic mailing list which is mostly used for announcing geo-related packages. Would this deserve its own Geoscience domain or should we create a geo tag and announcements and questions would go to the general list.
I wish there was a way to delegate administration of subcategories, but there doesn’t seem to be. As more people use this system, they’ll automatically gain trust and moderation capabilities though, so I think this will start to resolve itself.
What’s wrong with low traffic categories, as long as they do have some?
Suppose that in the future a not-yet-existing subcategory acquires traffic withing an existing big category. How do we expect to ever make the split reasonable, if discouragement is our criterion? Cause the old big category will still have more traffic than any new one (as is the case with Usage anyway), so that discouragement will always stay in place. Or who will bother discovering and moving the past relevant threads from the chaotic big category to the new subcategory? Therefore we’ll get stuck to the few categories decided early. Is that desirable? Do you have a different positive scenario in mind? Or does current consensus fail the whole concept of categories?
If the subject of a category is worthy to subscribe to, then low traffic is a plus, not a minus!
The only encouragement to use a category is for it to exist. If users end up not categorizing their posts and moderators end up moving many threads all the time, why not give moderators more categories to work with, along with giving users more specialized subscription options?
Forums create new subcategories all the time without this problem? Subcategoeries get made with mod help for a bit moving relevant topics, and people quickly switch. Hasn’t that always worked pretty well?
What hasn’t worked is creating too many categories. Julia-jobs? Julia-math? I could keep listing. When it splits too far it makes the community look inactive and makes everyone use one thread in order to get someone to actually read everything: Julia-users. So we know that method doesn’t work correctly.
There have been 33 posts in Julia-users with the word “audio” in its whole run. The majority of that is picking up posts which just have the word audio in it. JuliaAudio is an org, but it’s a small org without much usage yet (at least judging by the admittedly bad indicator of stars).
Telling people to post their 4 posts a year in such a small category is a good way to have no one read it, which defeats the purpose of the category. If every 4-5 posts per year topic is a separate category, there will be tons of categories most people won’t read, so most people won’t respond in, and so everyone will just post in the main forum. They need to be at least populated enough to be useful.
I think the solution that is going to work best for now is that we don’t over design what categories we have, but rather use tags to classify topics. If a certain domain/topic arises often enough we then create a category for those posts.
Also if a community wants to move to discourse e.g. julia-stats, or JuliaGPU, we can create a specific subcategory.
What you describe is valid only for small communities, which may easily switch and may indeed look inactive at times. It’s the exact opposite with big ones. A big community is very hard to switch and a bunch of low traffic forums cannot trick anyone into believing that the overall community is inactive. Unless someone hopes for Julia to stay small, it would be wise to look at the rate overall Julia traffic increases, even while currently being fragmented. Discourse cannot worsen things, only improve them. But as traffic increases (and Discourse will increase traffic even by itself), so does noise. Low noise is different from silence and can be a blessing.
And having people read what they don’t want to read (i.e. noise) defeats the purpose of the categories as well. We don’t tell people anything particular, cause chances are that people won’t listen anyway. Categories are mostly for subscriptions and proper movement of threads will be performed by moderators. There won’t be a single category without at least one moderator subscribed to it, so every single post will get read. Of course reading doesn’t guarantee a good answer, but that’s true for the main forum too.
Moreover, what most people miss is that Discourse is also a wiki. Can you imagine a wiki without specialized categories? Or should we defeat the purpose of wikis too? I’m afraid that most people jumped into the new platform thinking of it as just a little better than google-groups, without reading the features of the new platform nor the full reasoning for and vision behind the migration.