Improvement in the setting of the user directory

[N-thony]

Fixes #9467

Thanks @lilyclements for updating instatExtra::set_library_paths in IDEMSInternational/instatExtras#65
@rdstern, this will make sure that the user directory is created if not yet and set it as a primary library for packages before running the code to attach the packages in Rsetup.R. Now, when you update a package which goes to the user directory, it will be attached after restarting R-Instat.

[rdstern]

@N-thony not sure how and what to test here?
I started with sessionInfo() and found still the old versions were attached.

But I can't anymore load our packages. it says:

Should I be doing things in a different way?

[N-thony]

@N-thony not sure how and what to test here? I started with sessionInfo() and found still the old versions were attached.

But I can't anymore load our packages. it says:

Should I be doing things in a different way?

@rdstern @lilyclements, this is what I get when trying to install the databook package running my master branch.

[lilyclements]

Hi @N-thony I am going to investigate this. It's been working for @rdstern, but for me I've been getting this issue too. I think perhaps the devtools function has changed so our codes for 1, 2, ..., etc have changed
(We would say "If you get this warning, give this output", and for me it's been wrong for a few weeks now, but since @rdstern hasn't had an issue I've left it to prioritise other more urgent things before investigating). In the meantime can you install by devtools::install_github("IDEMSInternational/databook", update = "never")

[N-thony]

@lilyclements I also want to report something that seems a bit odd. The instatExtra::set_library_paths function was recently changed after the release of R-Instat version 0.8.3. However, when I run R-Instat 0.8.3, it appears to be using the latest version of instatExtra::set_library_paths instead of the one that was available at the time of the release. Am I missing something?

[N-thony]

@N-thony not sure how and what to test here? I started with sessionInfo() and found still the old versions were attached.
But I can't anymore load our packages. it says:

Should I be doing things in a different way?

@rdstern @lilyclements, this is what I get when trying to install the databook package running my master branch.

I got the same thing in R-Instat version 0.8.3 and this was due to an extra space, after removing the extra space, it seems to work

so, I have decided to install the databook

after installing, I was hoping to see the databook directory but instead I got the following packages

[lilyclements]

@N-thony great - glad it worked in the end. I'm not sure why the databook directory isn't there though

[N-thony]

@N-thony great - glad it worked in the end. I'm not sure why the databook directory isn't there though

@lilyclements I tried again, I still don't get the new databook in the user directory as done in R-Instat 0.8.2. I'm not sure if this issue is only on my side or @rdstern is facing the issue as well.

[rdstern]

@N-thony I am not sure if the following observation helps.
On the merged version (i.e. not this branch) when I go into R-Instat I run the first line of the following:

sessionInfo()
detach("package:instatCalculations", unload=TRUE)
library(instatCalculations)
detach("package:instatClimatic", unload=TRUE)
library(instatClimatic)
detach("package:instatExtras", unload=TRUE)
library(instatExtras)
detach("package:databook", unload=TRUE)
library(databook)

The sessionInfo() tells me that I have the original versions of the 4 Instat packages, i.e. not @lilyclements updated ones.
So I run the remaining lines, to detach and then attach the packages again. Then run the sessionInfo() again and I get the new versions, e.g. version 0.2.2 of instatExtras rather than the original 0.2.1.

Note I am not installing anything new, just changing the attached version.

When I try the same with your branch then:
a) The first sessionInfo() is the same as the merged version.
b) And the second sessionInfo() is unchanged from the first.
So, have you got the default (first directory) the wrong way round?

a) The

[N-thony]

@rdstern, I think this will be tested well in the release version, unless @ChrisMarsh82 has another opinion on this.