fread devel error

[arunsrinivasan]

Here's the link to zipped version of file (50MB, unzips to ~620MB).

require(data.table)
system.time(ans <- fread("~/Downloads/example.csv"))
# Read 374058 rows x 4 columns from 0.580GB file in 00:01.874 wall clock time (can be slowed down by any other open apps even if seemingly idle)
# Error in fread("example.csv") : 
#   Jump 12 did not end exactly where jump 13 found its first good line start. end(0x10b4f26ef)<<"3","8","7","2013/11/18 18:18:03">> != start(prev+35)<<"3","8","7","2013/11/18 18:19:02">>
# Timing stopped at: 4.473 0.247 1.879 

with verbose=TRUE:

None of the 1 'na.strings' are numeric (such as '-9999') which is normal and best for performance.
Input contains no \n. Taking this to be a filename to open
File opened, filesize is 0.579656 GB.
Memory mapping ... ok
Detected eol as \r\n (CRLF) in that order, the Windows standard.
Positioned on line 1 starting: <<"station_id","bikes_available">>
Detecting sep ...
  sep==','(ascii 44)  with 101 lines of 4 fields using quote rule 0
Detected 4 columns on line 1. This line is either column names or first data row (first 30 chars): <<"station_id","bikes_available">>
All the fields on line 1 are character fields. Treating as the column names.
Number of sampling jump points  = 11 because 3600 startSize * 10 NJUMPS * 2 = 72000 <= 622401253 bytes from line 2 to eof
Type codes (jump 00)    : 4444  Quote rule 0
Type codes (jump 10)    : 4444  Quote rule 0
=====
 Sampled 1031 rows (handled \n inside quoted fields) at 11 jump points including middle and very end
 Bytes from first data row on line 2 to the end of last row: 622401253
 Line length: mean=36.75 sd=0.75 min=36 max=38
 Estimated nrow: 622401253 / 36.75 = 16936648
 Initial alloc = 18630313 rows (16936648 + 10%) using bytes/max(mean-2*sd,min) clamped between [1.1*estn, 2.0*estn]
=====
Type codes (colClasses) : 4444
Type codes (drop|select): 4444
Allocating 4 column slots (4 - 0 dropped)
Reading 596 chunks of 0.996MB (28417 rows) using 4 threads
Read 374058 rows x 4 columns from 0.580GB file in 00:02.029 wall clock time (can be slowed down by any other open apps even if seemingly idle)
Thread buffers were grown 0 times (if all 4 threads each grew once, this figure would be 4)
Error in fread("example.csv", verbose = TRUE) : 
  Jump 12 did not end exactly where jump 13 found its first good line start. end(0x10b4f26ef)<<"3","8","7","2013/11/18 18:18:03">> != start(prev+35)<<"3","8","7","2013/11/18 18:19:02">>

This file was working fine a couple of commits ago but was thrice as slow (10s on CRAN version vs 32s on 4 threads). And now it errors.

Aside: also note that the columns are all read as character, even though the first three columns could/should be integers. But this happens even on CRAN version. I haven't had to time to check why this happens.