Open bbolker opened 2 months ago
Would be great to get this fixed. The older packages on CRAN have the same dependency. Very many thanks to the package authors if you are able to give this some attention.
This is really an "upstream" problem with the prediction
package, i.e. you should comment here or here. I did reach the maintainer of prediction
by e-mail, they said they were working on it.
In the meantime you should be able to use either remotes::install_github
to install first from leeper/prediction
, then leeper/margins
, then fsolt/dotwhisker
, or use remotes::install_version()
to install the last-archived versions of the corresponding packages from CRAN ...
e.g.
for (p in c("leeper/prediction", "leeper/margins", "fsolt/dotwhisker")) {
remotes::install_github(p)
}
I don't think any of these packages have compiled code, so you shouldn't even need to have developer tools installed ...
Thank you, that's very helpful - I had already tried remotes::install_version() but it failed due to the missing margins package. Will give these instructions a try.
Thank you, that's very helpful - I had already tried remotes::install_version() but it failed due to the missing margins package. Will give these instructions a try.
Just submit the new version to CRAN. Have a try to install from there, @LucieCBurgess . Let me know if there's any problem.
Sorry for not addressing this issue sooner...
Apparently, I missed the last email from Thomas Leeper with Ben ... My bad.
And sorry, @LucieCBurgess, it looks like we need to wait for Thomas to update the upstream package margin
to restore the package.
If he still cannot release the new margin
in a month or so, I will figure out how to switch to margineffect
-based function as Ben suggested.
In private e-mail Thomas said he had been busy with personal stuff but was hoping to re-submit margins
shortly.
@bbolker Thanks for you advise, remotes::install_github('fsolt/dotwhisker')
is just fine for me to install dotwhisker(Version 0.8.99)
The CRAN page says:
However, I can't for the life of me see where this dependence could come from. Neither theDESCRIPTION
file of the last-archived version on CRAN, nor the version on github, seem to show any signs of importingprediction
... ?? I suppose there's a possibility of a recursive dependence, but in that case why wouldn't the direct dependence be reported?Thanks to
r-package-devel
, I now know that this is through the dependence on themargins
package. Short of (1) removing this functionality, (2) switching to the more activemarginaleffects
package, I don't see a fix other than waiting for the maintainer of theprediction
package to resubmit to CRAN ...