revolutionanalytics / iterators Goto Github PK

This might be a misunderstanding of what the parameter is supposed to do, but I am attempting to use the chunksize parameter when creating an iter object and it is not producing what I would expect the output to be.

Example code:

out = foreach(lTmp = iterators::iter(1:100, chunksize = 15L)) %do% {
  print(length(lTmp))
}

I would expect the output to be 15, 15, 15, 15, 15, 15, 10, but the actual output is 1 repeated 100 times.

If I am misinterpreting, how difficult would it be to add an equivalent parameter to achieve that? I would be happy to fork the repo and put together a pull request if you could point me in the right direction.

Hello, I would like to call to your attention this issue on SO:
https://stackoverflow.com/a/70240075/2203727

Because RStudio gathers metadata for the "Environment" pane, the iterator is consumed by this process and results in faulty execution in the console. The reporter, Mikael Jagan, says that implementing a str function or refactoring the as.list function may resolve this issue.

In the meantime, turning on the "Manual Refresh Only" setting works around the issue.

By default, RStudio silently runs str(x) whenever you assign a value to the name x in the global environment, in order to populate the Environment pane. As a result,

library("iterators")
x <- iter(1:3)
x$state$i

returns 4L instead of 0L. (str.default evaluates vapply(x, typeof, "") and vapply evaluates as.list.iter(x).) The issue would be resolved if you wrote your own method for str avoiding as.list.iter, or if you moved the functionality of as.list.iter into a different function altogether so that str.default uses as.list.default.

This issue was brought to my attention by a question on Stack Overflow, here.