Upgrade to R 4.0.0?

Paul Evans

unread,

Apr 28, 2020, 5:42:50 PM4/28/20

to computationalstylistics

Hello All,

I'm wondering if it's advisable to upgrade to R 4.0.0 now that it's out, or to hold at R 3.6.3 for the time being. I'm only concerned about whether, and how well, stylo will work in 4.0.0. Most of my work is in Python, and I only use R as an environment in which to run stylo.

Thanks,

Paul Evans

Maciej Eder

unread,

Apr 29, 2020, 1:14:15 PM4/29/20

to computationalstylistics

Hi Paul,

a short answer might be: no need to upgrade, but YES, it is safe to do so.

A slightly longer answer: the version available on CRAN (i.e. 0.7.2) has been prepared to work with the new R 4.0.0. Today, I installed the 4.0.0 on my Mac, and tested 'stylo' once more. It worked OK. I've been still using R 3.3.2 on my Linux (Debian 9) machine and the new 'stylo' works with that one, too.

A more or less comprehensive answer: at CRAN, they've been discussing for several years about changing the way in which read.table(), as.table() and similar functions should handle strings. More on that here: https://developer.r-project.org/Blog/public/2020/02/16/stringsasfactors/index.html

Finally, they decided to switch once and forever. It affected dozens of packages, fortunately 'stylo' has always been resistant to this particular switch. I still have to screen the code, but as far as I'm concerned, all the tables that 'stylo' invokes, contain but numeric values. So, they won't be affected by the issue.

With this all said: should you encounter any strange behavior, please drop a line! I'll try to fix it as soon as possible.

All the best,

Maciej

David Hoover

unread,

May 5, 2020, 5:22:49 PM5/5/20

to Maciej Eder, computationalstylistics

Hi Guys,

I upgraded to R 4.0 a few days ago and found that Rolling.Classify no longer worked with SVM as the classifier. Delta and NSC work fine. Strange. I uninstalled R 4.0 and went back to 3.6X and all OK.

Peculiar,

David

--
David L. Hoover, Professor of English NYU Eng. Dept. 212-998-8832
https://wp.nyu.edu/davidlhoover/

“There is no nonsense so arrant that it cannot be made the creed

of the vast majority by adequate governmental action.”

-- Bertrand Russell, “An Outline of Intellectual Rubbish”

--
You received this message because you are subscribed to the Google Groups "computationalstylistics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to computationalstyl...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/computationalstylistics/11b91711-5b6c-45ad-80f8-c40fa10994e5%40googlegroups.com.

Maciej Eder

unread,

May 5, 2020, 5:53:30 PM5/5/20

to David L. Hoover, computationalstylistics

Hi David,

thanks so much for the feedback! I haven't noticed it so far, but I definitely have encountered a strange warning when using SVM with this new release of R. Warning message rather than an error, but still. Something to carefully inspect, for sure. I'll keep you posted about any progress on that: stay tuned!