Console doesn't respond to commands intermitently?

[dblodgett-usgs]

I know I'm not crazy, but I have no idea how to reproduce this. From time to time, when I enter a command in the console, I just get a blank new line -- like the command was never dispatched to R. It seems like it's when I ctrl-c to cancel out of what I was doing but now, when I try that, I can't reproduce. Is this something people have already reported and I can't find in the issues or should I continue to try and reproduce the issue so I can report it in a helpful way?

Here -- I reproduced it.

That targets dynamic branch name doesn't exist.

I typed

one <- 1 in the console, hit enter, pasted some random text, ctrl-c, printed the value of one, then tried to print a nonexistent variable. It just gave me a blank line.

This has happened to me a bunch where I do something that SHOULD work but it just does nothing like I never registered a command.

[juliasilge]

Hmmmmm, I don't think we have seen reports like this! 😱 I just tried to reproduce this behavior but could not locally; the nonexistent variable correctly errored with "object not found".

Can you set the log level for the R kernel to "trace":

And then the next time this happens to you, can you check out your kernel and console logs (in Output):

The console logs might be especially interesting, because that is where the logs show you the messages back from and forth from the console to R itself.

[dblodgett-usgs]

a) tab to accept the suggestion.
b) enter to try and evaluate the non-existent test console entry.

For the sake of completeness, I used ctrl+c to cancel the print command rather than evaluate it.

Trying to reproduce it again, I can do:

1. ctrl+f10 to restart the session
2. prin (observe suggestions)
3. tab to accept suggestion of princomp
4. ctrl+c to cancel the command
5. test (observe suggestions)
6. enter to accept testthat::
7. backspace a few times
8. enter to try and evaluate

And I get the behavior observed.

Fooling around with it -- it seems that I have to autocomplete a package (testthat::). I also have to either backspace or just left-arrow a couple times at step 7 to get it to happen.

So I can reproduce it with the above steps where I use any() and dplyr:: for example. It seems that I have to either backspace some of the autcompleted package or just move the cursor into the middle of the package name string before the enter keystroke as well.

[dblodgett-usgs]

I want to document this here as I have never noticed issues from these firewall errors -- it seems like whatever this is doesn't actually have any affect -- but I increasingly wonder if this setting that our IT (US Geological Survey) is to blame for something.

[juliasilge]

Ah, it may in fact be something about the firewall and ark, yep. Are you able to find messages in the console logs like I outline here, and compare what messages are there when the console is behaving correctly vs. not? You'll find these logs in the output panel. It would be helpful to know which messages are/are not getting sent back and forth from the console to ark.

[juliasilge]

Also, I don't suppose it gives you any helpful info if you click "Show more" in that Windows security dialog? What would be really helpful would be to figure out what messages are not getting through to/from ark to the console.

It would also likely be helpful to look at the kernel logs (the ark side) to see what message are arriving/leaving there.

[dblodgett-usgs]

Not ignoring here. Just off focused on other things for a bit. I'll look into this when I free up some cycles.