update on cran before 30/10

[DominiqueMakowski]

Dear maintainer,

Please see the problems shown on
https://cran.r-project.org/web/checks/check_results_bayestestR.html.

Please correct before 2021-10-19 to safely retain your package on CRAN.

_{which is also the date of my birthday}

[strengejacke]

^{which is also the date of my birthday}

🥳

[bwiernik]

Ain't no party like a CRAN deadline party.

[strengejacke]

@mattansb @DominiqueMakowski There a few tests failing on CRAN (https://www.r-project.org/nosvn/R.check/r-devel-windows-x86_64/bayestestR-00check.html), but some more here on GitHub. Are you able to fix these in time?

[mattansb]

I have a PR (#474) that I will not merge yet, as it introduces some more complications.

I don't know what the CRAN errors are about - looks like Dom's domain.

As for the error on the main branch - do you know where the error is happening? I can't figure it out from the log at https://github.com/easystats/bayestestR/runs/3895314216?check_suite_focus=true

[strengejacke]

This is the current error log:
https://github.com/easystats/bayestestR/runs/3913495285

[strengejacke]

Looks like some recent changes in insight that cause problems in bayestestR need to be fixed...

[strengejacke]

@mattansb I'm not sure, there seem to be some issues with unupdate(): https://github.com/easystats/bayestestR/runs/3913495285#step:10:1014

[mattansb]

Weird... Seems like cmdstanr needs to be added to Suggests 🤷

[DominiqueMakowski]

Hello 👋

need to be fixed...

you mean fixing as in commenting out the failing tests 😅?

[mattansb]

The current errors seem to have something to do with BRMS? Taking too long? I'm not quite sure how to read the logs...

[strengejacke]

══ Failed tests ════════════════════════════════════════════════════════════════
── Error (test-bayesfactor_parameters.R:85:7): bayesfactor_parameters BRMS ─────
Error: Error in `unupdate.brmsfit(brms_mixed_6)`: Error : Please install the 'cmdstanr' package.

Backtrace:
    █
 1. ├─bayestestR::unupdate(brms_mixed_6) test-bayesfactor_parameters.R:85:6
 2. └─bayestestR:::unupdate.brmsfit(brms_mixed_6)
── Failure (test-check_prior.R:36:3): (code run outside of `test_that()`) ──────
check_prior(model2)$Prior_Quality (`actual`) not equal to c(...) (`expected`).

    actual          | expected           
[1] "uninformative" | "uninformative" [1]
[2] "informative"   - "uninformative" [2]
[3] "informative"   | "informative"   [3]
[4] "uninformative" | "uninformative" [4]
[5] "uninformative" | "uninformative" [5]
── Failure (test-check_prior.R:44:3): (code run outside of `test_that()`) ──────
check_prior(model2, method = "lakeland")$Prior_Quality (`actual`) not equal to c(...) (`expected`).

    actual           | expected         
[1] "informative"    | "informative" [1]
[2] "misinformative" - "informative" [2]
[3] "informative"    | "informative" [3]
[4] "informative"    | "informative" [4]
[5] "informative"    | "informative" [5]
── Failure (test-describe_posterior.R:265:7): (code run outside of `test_that()`) ──
describe_posterior(ttestBF(mtcars$wt, mu = 3), ci = 0.95) (`actual`) not equal to structure(...) (`expected`).

actual vs expected
                  CI_high
- actual[1, ]   0.1557581
+ expected[1, ] 0.1792614

  `actual[[5]]`: 0.16
`expected[[5]]`: 0.18

[ FAIL 4 | WARN 23 | SKIP 3 | PASS 317 ]
Error: Error: Test failures

[mattansb]

That cmdstanr error is very weird... What that model insight::download_model("brms_mixed_6") fit with cmdstanr???

Anyway, you can comment that chunk out for now (sorry, I don't have time to get to this.......)

[strengejacke]

ok, @DominiqueMakowski, you may sent to win-builder and then to CRAN? Looks like all other issues have been resolved...

[DominiqueMakowski]

https://win-builder.r-project.org/VvnwB1Cyt7MJ
https://win-builder.r-project.org/vNdFyScOlN9
https://win-builder.r-project.org/yelYQqydBFDO

[strengejacke]

https://easystats.github.io/parameters/reference/ci.merMod.html is now https://easystats.github.io/parameters/reference/ci.default.html

The two failing tests are probably cmdstanr issues? I suggest skipping those on CRAN.

[strengejacke]

Oh, and you need to remove the remotes field from the description file :-)

[DominiqueMakowski]

Thanks, we see:

 URL: https://easystats.github.io/parameters/reference/ci.merMod.html
   From: man/ci.Rd
   Status: 404
   Message: Not Found

Please fix and resubmit.

[strengejacke]

Thanks, we see:

 URL: https://easystats.github.io/parameters/reference/ci.merMod.html
   From: man/ci.Rd
   Status: 404
   Message: Not Found

Please fix and resubmit.

See my comment (#469 (comment)) above:
https://easystats.github.io/parameters/reference/ci.merMod.html is now https://easystats.github.io/parameters/reference/ci.default.html

[DominiqueMakowski]

See my comment

Oops, sorry I am now an old man my memory and attention span are gone ^^

[strengejacke]

[IndrajeetPatil]

I am guessing the archival deadline was extended to 2021-11-02?

[DominiqueMakowski]

til the 31/10, but i submitted yesterday and I am now waiting for a reply about some dead links false positives

[IndrajeetPatil]

I would consider re-submitting (no need to change version number) because it can take days if not weeks before someone looks at the e-mail, and we don't want to push our luck.

[DominiqueMakowski]

yeah you're right, but because of the "possibly dead links" message as I didn't fix them i am worried that it will get back to me again without a manual inspection

[IndrajeetPatil]

If they are not too crucial, I'd say skip those URLs for now and resubmit.

[DominiqueMakowski]

let's do that, I'll resubmit today

[DominiqueMakowski]

cran contacted me 2 days ago after submission that it breaks some down downstream deps (see) I said we are aware and will fix asap but since then no news :/ (status is waiting in https://lockedata.github.io/cransays/articles/dashboard.html)

[strengejacke]

Strange, but why did this not occur during submission? Anyway, I have revised tests, forcing them to be skipped on CRAN, that's the safest way I'd say,

[IndrajeetPatil]

How did {see} update ever make it to CRAN?! This test should have failed during the submission process itself.

[IndrajeetPatil]

Strange indeed.

At any rate, the safest way to resolve this ASAP will then be:

• update see on CRAN?
• run revdepcheck for bayetestR again
• mention in cran_comments if you see any problems (no problems seen)
• resubmit bayestestR

[DominiqueMakowski]

Rchecks work in mysterious ways

[IndrajeetPatil]

I am running revdepcheck now and will make a PR in a bit

[strengejacke]

Note that updating see will completely break plotting for correlation.

[IndrajeetPatil]

How does every update in one package in our ecosystem breaks something in another package?! 😭

[IndrajeetPatil]

So bizarre, I see no issue with CRAN {see} in revdepcheck. I am guessing CRAN is running theirs on some other architecture.

-- CHECK ---------------------------------------------------------------------------- 18 packages --
√ coveffectsplot 0.0.9.1                 -- E: 0     | W: 0     | N: 1                              
√ correlation 0.7.1                      -- E: 0     | W: 0     | N: 0                              
√ datawizard 0.2.1                       -- E: 0     | W: 0     | N: 0                              
√ effectsize 0.5                         -- E: 0     | W: 0     | N: 0                              
√ emmeans 1.7.0                          -- E: 0     | W: 0     | N: 1                              
√ eiCompare 3.0.0                        -- E: 0     | W: 0     | N: 0                              
√ fbst 1.5                               -- E: 0     | W: 0     | N: 0                              
√ multifear 0.1.2                        -- E: 0     | W: 0     | N: 1                              
√ modelbased 0.7.0                       -- E: 0     | W: 0     | N: 1                              
√ neatStats 1.7.3                        -- E: 0     | W: 0     | N: 0                              
√ insight 0.14.5                         -- E: 0     | W: 0     | N: 0                                                                  
√ performance 0.8.0                      -- E: 0     | W: 0     | N: 0                                                                  
√ psycho 0.6.1                           -- E: 0     | W: 0     | N: 0                                                                  
√ report 0.4.0                           -- E: 0     | W: 0     | N: 0                                                                  
√ parameters 0.15.0                      -- E: 0     | W: 0     | N: 0                                                                  
√ sjstats 0.18.1                         -- E: 0     | W: 0     | N: 0                                                                  
√ see 0.6.8                              -- E: 0     | W: 0     | N: 0                                                                  
√ sjPlot 2.8.9                           -- E: 0     | W: 0     | N: 0                                                                  
OK: 18                                                                                            
BROKEN: 0

[strengejacke]

How does every update in one package in our ecosystem breaks something in another package?! 😭

This is due to the change in the underlying plotting API, namely the switch to visualization_recipe(). While this is a clever change, it's a big breaking change, so this stuf is expected and will hopefully be resolved in the next release cycle(s).

Or in other words: @DominiqueMakowski managed it to introduce ecosystem breaks in each single easystats package :-D

[IndrajeetPatil]

Dom giveth, and Dom taketh away...

[IndrajeetPatil]

I think we should ask for another extension for submission. This is all not going to be resolved by tomorrow.

What do you think?

[strengejacke]

We can simply submit bayestestR, because failing checks are not related to see. And then submit an update for see within the next two weeks.

[DominiqueMakowski]

On its way to cran

[strengejacke]

Interestingly, see does not yet fail on CRAN: https://cran.r-project.org/web/checks/check_results_see.html

Not sure if we will expect something else the next days?

[DominiqueMakowski]

I don't think there will be more breaking changes related to Vizrecipes as what led to breaks was the iterative moving of the function definition and related changes, but now that's stabilized so it should be okay. I don't know why but today I feel like the new dawn of a stable easytats has started 🌄

[strengejacke]

I don't think there will be more breaking changes related to Vizrecipes

No no, I was referring to the CRAN issue from revdep checks (breaking see), but still, as of today, CRAN checks for see are all ok. So the revdep-issue could have been a false positive.

[IndrajeetPatil]

Maybe we can submit {modelbased} now?

[DominiqueMakowski]

should we wait for see ?

[strengejacke]

Do we need to wait for see? And I'm not sure current see works with parameters.

[DominiqueMakowski]

well then ok i'll submit modelbased asap

[IndrajeetPatil]

Any update on the submission? Is it stuck at some stage?

[DominiqueMakowski]

stuck at the "didn't submit yet" stage ^^ but will move it to "submit now"

[strengejacke]

If you're talking about modelbased, are all relevant issues resolved and PRs merged? Is there any hurry, or should we see until all relevant stuff is implemented?

[DominiqueMakowski]

I don't think there's a hurry unless @IndrajeetPatil knows something I missed

[IndrajeetPatil]

No hurry. I just wanted the ecosystem to remain in sync.
The last release for all packages was in September/October, except {modelbased}.

[strengejacke]

I'd say following should at least be addressed before submission:

• Clarify expectation vs prediction, deprecate estimate_response() modelbased#152
• get rid of NOTE in CRAN checks modelbased#157
• visualisation_recipe with transformed responses  modelbased#151
• Include response in estimate_predicted() data frame modelbased#142

[DominiqueMakowski]