Serious issue on where R-Instat looks for R-packages?

[rdstern]

@N-thony and @ChrisMarsh82 I think we have uncovered the serious problem that is plaguing our export for Google buckets.

I write because if I am right, then I want all the praise, but @jkmusyoka and @lilyclements have been doing all the work!

The problem is that the export dialog fails on most machines and seemingly in different ways. It works perfectly on James and Lily's machines! It worked on my machine in Malawi, and now it did not.

Lily has (magically) diagnosed that it could be purrr that is the problem and I find that I have purrr 1.01, while it was updated to purrr 1.02 on 10 August. James has updated purrr and also rlang on my machine and now it works.

Now our main discovery. On Micah's machine she is using purrr 0.34 and it gives an error. James has deleted R-Instat purrr and checked and she is still using purrr 0.34!!!!

So we deduce that when R is installed already, then R-Instat appears to be looking first in the R packages and only afterwards does it look in the R-Instat ones?

That is at least an explanation of the different errors we get, that seem to be machine dependent.

Could you please check urgently - maybe even include Danny if you feel it is necessary?

In making the changes could you also please remember we are hoping in the future that R packages in R-Instat could be addedm witht needing to be an adminstrator. Maybe that could be added at the same time?

James may add!

[rdstern]

@ChrisMarsh82 I assume this is your sort of problem and wonder about your silence?

[rdstern]

Delighted to hear that @N-thony has been contacting @ChrisMarsh82 and a solution may be in sight. I assume earlier, that the packages were separate, and understand from Danny that this is a line or so that needs to be included in the installation procedure. So I hope we can revert to the previous situation.

I saw more evidence of the new problem and think I understand why it happened. I note two instances:
a) On Micah's machine in Zambia, she was the only one who could not do (I think) an inventory graph. The error message said she was using ggplot2 version 3.46 and needed at least 3.5. This satisfies my "theory" that the issue may have been around for quite a long time. It is not evident for anyone involved in development, because they use the R packages than - so they are kept up-to-date. It was not evident in West Africa, because I assume there was no R installed separately.
b) So it is a potential problem with non-developers who have R installed and particularly (I assume) if they have not used it for some time. I am not quite clear why the Exception occurs on some machines. This is the first time I have seen that. James managed to fix that on most machines, in Zambia, but not all. I am also getting it on Danny's machine - which is good, because we can check if the fix works there. I installed Version 0.7.16 which is the latest version on the web site, and that is working fine. I note it is more than a year old, and suggest we update to version 0.7.21 as soon as we can.
c) I have a list of a few items that we found during the workshop. But - this major problem apart - it held up far better than any previous version.

[rdstern]

@N-thony and @jkmusyoka I wonder if this stack-overflow question is any help?

I have anyway, just had a slightly confusing time with Danny's laptop. Earlier, when installing Version 0.7.20 it gave the dreaded exception error we had on various machines in Zambia. I was happy - in a way - because this could then test the new version. However, I have just tried again today and now it loads Version 0.7.20 just fine! I tried as administrator and then without. It seems fine now. I have downloaded and run Version 0.7.16 (since getting the 0.7.20 exception) and wonder if that had any influence?

That's a relief to me, because it is working pretty well, apart from this "feature". I go to Zimbabwe next week, and could demonstrate using this version if appropriate.

Then using the stackoverflow I now ran the following line in the script window:

.libPaths("C:/Program Files/R-Instat/static/R/library")

When I now type .libPaths( ) it now gives me:

[1] "C:/Program Files/R-Instat/static/R/library"       
[2] "C:/Program Files/R-Instat/0.7.20/static/R/library"

Maybe I should have used 0.7.20 when specifying the path?

And now a more interesting result that might help the fix. I went out of R-Instat and back in again (ordinary, not administrator). Then did the .libPaths() command as the first thing.

And I got [1] "C:/Program Files/R-Instat/0.7.16/static/R/library"
So, now I do think that my starting on Danny's machine, on which 0.7.8 was the latest. Then I got the exception. Then I needed R-Instat so I installed version 0.7.16. Now the exception error disappeared, and the library 0.7.20 is trying to use is the older 0.7.16 library.

@N-thony and @ChrisMarsh82 there must see some clues here to help solve the 0.7.20 library problem?

[rdstern]

@N-thony and @jkmusyoka and @ChrisMarsh82 I have just installed the latest version produced today.
I then typed .libPaths() it gave:

[1] "C:/Program Files/R-Instat/0.7.20a/static/R/library"

(I installed in 0.7.20a because I had already installed the previous version into 20 and didn't yet want to delete it.)

This is perfect. Just what it should be! Many thanks. I'll continue testing, but so far so good!

[N-thony]

@rdstern can we close this?