I have local installations build/install of packages that are also on CRAN.
When I updated to yesterday's build to this setting
version R version 3.5.1 (2018-07-02)
system x86_64, darwin15.6.0
ui RStudio (1.2.997)
language (EN)
collate en_US.UTF-8
tz America/New_York
date 2018-09-20
When i build vignettes what i see in the interactive build and previewing the created htmls in the inst/doc are not the same as when i access them through the packages window.
When are force remove all the packages from the libpaths and rebuild things are as expected in the help, but when building a pkgdown site the .libPath that it is looking at is in Frameworks which is doesnt have a copy of the install any longer.
What was the last version of RStudio you were using where everything looked fine? (Have you been running any of the previous preview releases / daily builds?)
Can you elaborate on what you mean, here? Are you setting custom library paths somewhere that are being lost / reverted?
i was on a previous daily build and it was working ok (not sure which one it was, if i can find an old reprex with it i'll update).
I am not setting any custom libpaths.
for now i can get it to work better by reversing the current default libpaths in an .Rprofile. then things seem to be working relatively ok, but the IDE gets lost in the new auto check for notifying if a package is missing in order to run a script (since it looks like that is looking in the default .libPaths()[1], which is now .libPaths()[2])
I see the hover working, but i am getting two values for an installed package (in this case pkgconfig) which i dont have a fork locally, i just installed it via install.packages
where one is in the user and the other in the system. this is also true for my locally built packages that shouldnt be in system (right?)
If you're still seeing package(s) being unexpected installed into that directory, I think we'll need a reproducible example so we can verify the behavior locally.