read_dta leads to immediate R crash

[jon-mellon]

When I try reading a .dta file into R using read_dta, it causes a complete crash of R.
The following minimal file causes this crash:
https://github.com/jon-mellon/mellonMisc/blob/master/data/panel_leader.dta

Details of my system are below. Note that I am using Rstudio version 0.98.1103
_
platform x86_64-w64-mingw32
arch x86_64
os mingw32
system x86_64, mingw32
status
major 3
minor 2.0
year 2015
month 04
day 16
svn rev 68180
language R
version.string R version 3.2.0 (2015-04-16)
nickname Full of Ingredients

Sys.info()
sysname release version nodename machine
"Windows" "7 x64" "build 9200" "JON-SURFACE" "x86-64"
login user effective_user
"Jon" "Jon" "Jon"

[jon-mellon]

Note that I suspect the issue is with this dataset, but it's not an issue that is causing any known problems with stata or SPSS, so I suspect many files out there may suffer from these issues too.

[hadley]

I've fixed the crash, but I'm not sure I'm getting the correct value.

What value do you see in stata for generalElectionVotePostOthW5 in row 41838?

[jon-mellon]

My file only has 41837 rows (excluding headers). If you mean row 41837 (the last row in the file), then running:

list generalElectionVotePostOthW5 if id==34803

gives

       | g~tOthW5 |
       |----------|
41837. |          |

[jon-mellon]

By the way, I was getting a related error when trying to read in the SPSS version of this file as well. Do you want me to open up a separate issue or will it be sorted by this one?

[hadley]

Ok so that variable uses both __NA__ and nothing to represent missing values?

The SPSS issue will probably be fixed as well, but please try it out with the dev version and let me know if it doesn't work

[jon-mellon]

Yes, the __NA__ items are created by the survey company, the blanks are the result of merging versions of the datasets (where the question wasn't asked to all respondents in all waves).