错误:包是由一个R版本安装的，有不同的内部；它需要重新安装

Question

我遇到了一个软件包安装问题，在那里，我完全不知道如何管理的情况。

install.packages("C:/Users/Oliver/Downloads/NPRED.zip", repos = NULL, type = "win.binary")
Installing package into ‘C:/Users/Oliver/Documents/R/win-library/3.5’
(as ‘lib’ is unspecified)
package ‘NPRED’ successfully unpacked and MD5 sums checked
library("NPRED", lib.loc="~/R/win-library/3.5")

*

> Error: package or namespace load failed for ‘NPRED’:
>      package ‘NPRED’ was installed by an R version with different internals; it needs to be reinstalled for use with this R version

*

有人能帮我解决那个问题吗？

非常感谢！干杯，奥利

Answer 1

# Find the currently installed packages, put this in a data frame
df_all_packages <- as.data.frame(installed.packages())

看看这个，检查哪个主要和次要的R版本在那里。

View(df_all_packages)
table(df_all_packages$Built) # how many packages built under each version of R

下面的示例假设您使用R4并希望基于R3.6升级一些包。你应该根据你自己的情况调整一下。

# Find older package versions, make a list with packages which are of older version and you want to update
df_packages_to_upgrade <- as.data.frame(subset(df_all_packages, (Built %in% c('3.6.1', '3.6.2','3.6.3'))))
# Or you can prefer to base the filtering on those you want to keep, if so uncomment next line:
# df_packages_to_upgrade <- as.data.frame(subset(df_all_packages, !(Built %in% c('4.0.1', '4.0.2','4.0.3'))))

# Finally re-install the packages :
install.packages(df_packages_to_upgrade$Package)

有时候，上面提到的都会失败。特定包的依赖项可能会发生更改，这可能很难跟踪；这可能涉及到棘手的OS级依赖关系。在一种情况下，我不得不：

# At bash command line, as root:
yum install jq-devel protobuf-devel libcurl-devel openssl-devel geos-devel gdal gdal-devel proj-devel proj-nad proj-epsg libjpeg-turbo*
# use 'apt-get install' on Debian-based
# Then, in R (as linux super-user), remove all old packages - but before, make a copy-paste of the list, just in case
# The list:
old_packages_removed <- rownames(df_packages_to_upgrade)

remove.packages(df_packages_to_upgrade$Package)
# Then I installed the 2 or 3 packages I really needed, which ended up installing 50 packages (most of the previously un-installed ones)
# You could try, instead:
install.packages(old_packages_removed)

这为我解决了这个问题。