[BUG] fread error on particular csv in latest development build

[yaakovfeldman]

I have several large csv files. Using the latest development build (windows binary from AppVeyor), fread fails on one of them with error:

Error in fread("myfile.csv") : A row outside the large sample was i) either too short and fill=FALSE ii) too long or iii) a jump batch didn't end properly. This message will be made specific before release to CRAN.

This is after the commit 'fread parallel' 0b3cd65

The previous commit and earlier still work fine, so something must have changed in this commit.

The csv has 89205 observations of 9 variables. Unfortunately, it is sensitive. What can I try to help make a reproducible example?

EDIT: anonymised csv in comment below

[MichaelChirico]

is it possible to replace all of the sensitive fields with ### or some other anonymizer?

…

On Mar 21, 2017 8:32 AM, "yaakovfeldman" ***@***.***> wrote: I have several large csv files. Using the latest development build (windows binary from AppVeyor), fread fails on one of them with error: Error in fread("myfile.csv") : A row outside the large sample was i) either too short and fill=FALSE ii) too long or iii) a jump batch didn't end properly. This message will be made specific before release to CRAN. This is after the commit 'fread parallel' 0b3cd65 <0b3cd65> The previous commit and earlier still work fine, so something must have changed in this commit. The csv has 89205 observations of 9 variables. Unfortunately, it is sensitive. What can I try to help make a reproducible example? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2070>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AHQQdUwIDEGuAg98CpfljGpsW-oGOFKIks5rn8NJgaJpZM4Mjur_> .

[yaakovfeldman]

I now have a reproducible example! (Anonymising didn't work earlier)
csv.zip

I get the error above with
fread(myfile.csv)

[mattdowle]

Reproduced - thanks.

[mattdowle]

Many thanks for providing this file. Works for me now. Please open a new issue if you find anything else.