I think opening up the developers category has been brought up a few times. How about we vote until the next Stan meeting (Thursday) and we can take action then?
Just FYI, the voting results are public.
Open developers category (everyone can read and write)
Leave developers category as-is (everyone can read, only approved members can write)
If we decide to open up the category, I think these are the steps we should take in Discourse:
Add a āStan Core Development Teamā group; add the people listed as active developers on mc-stan.org to that group
Rename the Developers only category to āStan Core Internal Discussionsā or something similar (I think we want to un-nest it from the Developers category); set permissions so that itās read/write for this new group
Wasnāt there a conversation about this on here about 2 weeks ago?
Has there been a problem other than Bobās Stan 3 proposal thread, which definitely shouldnāt been more open because not all stakeholders could comment.
Edit to say: A private Dev-only category is useful. Itās rarely used, but all the issues Iāve seen raised in it couldnāt be raised on here in public. And a post here minimises email traffic/gets to everyone. Losing it would be inconvenient.
(For those who canāt see the category, it does not contain development discussions.)
Edit 2: Changing permissions on a private category to make it public should involve permission from everyone who posted in it and a careful check nothing sensitive was posted there. It would be better to start a new category and let the old one die.
Yes, but no resolution. It was part of a thread. Iām just laying this out with the options and a path forward if we take the option to change permissions.
+1. To all of this! Exactly. Itās rarely used, but when used, itās appropriate. And itās not for development discussion.
Sorry, I must not have been clear.
Iām not suggesting opening up the private category. I am suggesting that the group that can access it should b restricted to only the people listed on the website and that it remains private. We currently give people who ask access to the developers category, which automatically grants access to the private subcategory.
Currently we have a private dev category and a public dev category that users canāt post to. Is the goal here to make the public dev category one that users can post to? That seems fine. I donāt know why that involves the shuffle you suggest below but thatās ok.
That was the original intention, but in practice, not quite.
The access to that subcategory is defined by the people within the ādevelopersā group. As people request access to participate on the developers category, we just grant them access (which gives access to both). The ādeveleoprsā group is actually larger than just the core devs.
Stan (code) development! Stuff like implementation details for Math, Stan, or the interfaces, code design decisions, new features and bugs, etc. The stuff that used to be on stan-dev.
The stuff that used to be in stan-users is everywhere else. That includes users reporting installation issues, modeling questions, most modeling implementation discussions. Thereās a gray area in between that can go in either place.
Make sense? I think the stricter separation and purpose got lost when the users list became a number of categories in discourse.
Thank you for doing this. My background is in software development. As someone who is mostly lurking and new to stan, I canāt give much back in the discussion of models. But I can help with āDevelopersā questions.
My first day on the site a senior developer was trying to figure out how to use discourse. So I was excited to bitch in. I wrote up a long response answering each of his questions from my personal experience using discourse in other communities. Then I wanted to ensure I shode good faith. So I found documentation for each point. (fyi, There is surprisingly lital on what version of markdown is used. I eventually linked to the source code.) Only to discover that I could not post it as I was not a Developer.
There was an open question on whether pystan should interop with pandas. But now I know not to try and respond.
There is an ongoing discussion on testing, how much, when and where. I lurk/follow on the rust-lang infrastructure team. I thout Iād suggest reading some of the thoughts the rust-lang devs have on this, especially the āNot Rocket Science Ruleā.
I just wanted to share some anecdotes from my experience. I would have posted earlier, but the dragon eats its own tail; the discussion of being more inclusive was not itself inclusive.
So thank you for opening this up. I understand that there are tradeoffs. I hope the experiment goes well.
(I take this out of the context and answer specifically to these lines)
Hi, comments are always welcome and any discussion considering PyStan should be non-hostile. So even if the discussion has āendedā, please give your comment on subject.Z