CmdStanR 1.0 release

[jgabry]

It came up in on the forum that CmdStanR not being at v1.0 gives the impression that the API isn't considered stable yet.

@andrjohns @SteveBronder what do you think about going ahead with the 1.0 release?

Is there anything you want to get in before we do that? I think it would be good to get #1022 merged, although I guess we could do 1.0 without it and merge it after. Anything else?

[andrjohns]

On board with 1.0 soon! One thing we did chat about before for a CRAN release was removing the backwards compatibility so that we're starting with a simpler codebase, but not sure how essential that is

One other minor thing that I want to test out and open up a PR for is extending #1054 for RTools{40,42,43}. From my testing in the original Math PR it was compatible with the Math stuff, but just want to double-check with cmdstan/cmdstanr.

If that has no issues, and we can restrict compatibility to cmdstan >= 2.35.0, then we could also remove all of the custom toolchain installation/handling on windows

[jgabry]

On board with 1.0 soon! One thing we did chat about before for a CRAN release was removing the backwards compatibility so that we're starting with a simpler codebase, but not sure how essential that is

Oh, good point. We should do this. I can work on it.

One other minor thing that I want to test out and open up a PR for is extending #1054 for RTools{40,42,43}. From my testing in the stan-dev/math#2999 it was compatible with the Math stuff, but just want to double-check with cmdstan/cmdstanr.

Sounds good.

If that has no issues, and we can restrict compatibility to cmdstan >= 2.35.0, then we could also remove all of the custom toolchain installation/handling on windows

Yeah that would be great!

[SteveBronder]

A few things I'd like to have/try before we do a 1.0 release

• Improve Handling of cpp_options #1022 (@katrinabrock I really apologize for my tardiness on that)
• Instead of starting N cmdstan processes I'd really like to give using cmdstan internal multithreading across chains a go. It should be a nice speed update for users. We can always backup to multiple cmdstan instances if need be
• Feature/reduce mem pathfinder stan#3325 adds a new column to pathfinder's output so we will need to account for that in the code

[jgabry]

A few things I'd like to have/try before we do a 1.0 release

• Improve Handling of cpp_options #1022 (@katrinabrock I really apologize for my tardiness on that)

Agree this would be great to get in.

• Instead of starting N cmdstan processes I'd really like to give using cmdstan internal multithreading across chains a go. It should be a nice speed update for users. We can always backup to multiple cmdstan instances if need be

How much work do you think it would be to do this?

• Feature/reduce mem pathfinder stan#3325 adds a new column to pathfinder's output so we will need to account for that in the code

Sounds good

[SteveBronder]

How much work do you think it would be to do this?

A lot. Let's have that be a "would be nice" but not totally necessary.

[katrinabrock]

Is making the package CRAN-ready a goal eventually? Or is it not intended to be on CRAN?

[andrjohns]

Yep, CRAN is a goal

[katrinabrock]

Cool! I feel like that will help a lot because then you can look at Reverse Depends to see how people are actually using it. (Far from exhaustive, of course, but better than not having that info.)