Closed acksmaggart closed 3 years ago
As soon as I can get the tests to run I'll mark is as "Ready for review". Rserve on my machine must have been installed differently than on yours because the testtools.py
module isn't detecting the path correctly.
Hi Max, did you get any progress on this issue? R
Hey, @ralhei . Sorry for the wait. I have made the two changes you suggested and re-based onto master
, let me know if anything still looks squirrelly. And thanks for the package!
Hi Max, thanks for your changes. However it is never a good idea to merge changes from master into a working tree because the history tree really gets messed up. So please (in the future) always 'rebase' your working tree onto master and avoid a merge. I'll accept your change anyway, just to get ahead with everything. Cheers, Ralph
Sorry, changed my mind, because changes from Chad and myself would be part of the pull request.
Before running the next command please create a patch for your changes, just to be safe ;-)
Then the command to get rid of the merge-commit in your history is to run git rebase --onto 62c4d36e^ 62c4d36e HEAD
Finally push (possibly with --force) again. We hopefully then have a clean pull request.
Hi Max, in order to get this PR finally done I've finally merge your code myself in the proposed way. Due to that I will now close the PR. Thanks a lot for your contribution.
I know I'm wading into controversial water here, but there are cases where R's "NA" value is being converted into something unexpected in python. Specifically:
vector
s that containedcharacter
values andNA
were causingUnicodeDecodeErrors
as they were being parsed intonumpy
arrays. The change inpyRserve/misc.py
now has the parser checking for R'sNA_character_
specifically and converting it to Python'sNone
.vector
that contains booleans andNA
produces a pythonnumpy
array of integers andNone
. TheNone
is expected, but the integers are a surprise. The change inpyRserve/rparser.py
produces a numpy array of booleans andNone
.