Cmdstanr: Can't devtools::install_github("stan-dev/cmdstanr")

[nerutenbeck]

I'm trying to install cmdstanr from GitHub via devtools::install_github("stan-dev/cmdstanr")

Expected behavior

Successful package install followed by the need to call cmdstanr::install_cmdstan(), which installs CmdStan and sets the path.

Error

Installation fails with the following:

✔  checking for file ‘/tmp/RtmpRH3cgr/remotesb1b5b912816/stan-dev-cmdstanr-edf9bad/DESCRIPTION’ ...
─  preparing ‘cmdstanr’:
✔  checking DESCRIPTION meta-information ...
─  checking for LF line-endings in source and make files and shell scripts
─  checking for empty or unneeded directories
─  building ‘cmdstanr_0.1.1.tar.gz’

Installing package into ‘/usr/local/lib/R/site-library’
(as ‘lib’ is unspecified)
* installing *source* package ‘cmdstanr’ ...
** using staged installation
** R
** inst
** byte-compile and prepare package for lazy loading
** help
*** installing help indices
*** copying figures
** building package indices
** installing vignettes
** testing if installed package can be loaded from temporary location
Error: package or namespace load failed for ‘cmdstanr’:
 .onAttach failed in attachNamespace() for 'cmdstanr', details:
  call: NULL
  error: CmdStan path has not been set yet. See ?set_cmdstan_path.
Error: loading failed
Execution halted
ERROR: loading failed
* removing ‘/usr/local/lib/R/site-library/cmdstanr’
Error: Failed to install 'cmdstanr' from GitHub:
  (converted from warning) installation of package ‘/tmp/RtmpRH3cgr/fileb1b12559732/cmdstanr_0.1.1.tar.gz’ had non-zero exit status

To Reproduce
On a fresh machine image without the package or CmdStan installed / path set run devtools::install_github(...)

Operating system
Amazon Linux 2 AMI

CmdStanR version number
GitHub master

Additional context

I can still install via install.packages("cmdstanr", repos = ...)

[nerutenbeck]

Could be something in #273? I successfully installed with this method on a fresh machine image earlier this week, prior to that PR.

[rok-cesnovar]

Indeed, the problem is that check fails if there is no "current version" to check against. Let me see if there is an easy fix for that. Otherwise I will revert that PR and we will reapply a fixed version of that.

[jgabry]

I think I have a fix for this. Can open a PR in a few min

[jgabry]

@nerutenbeck Thanks for reporting this. This is definitely due to a mistake I made in #273. Should be fixed by #275.