[issue24666] Buffered I/O does not take file position into account when reading blocks

[Eric Pruitt]

New submission from Eric Pruitt:

When buffering data from a file, the buffered I/O does not take into account the current file descriptor position. In the following example, I open a file and seek forward 1,000 bytes:

>>> f = open("test-file", "rb")
>>> f.seek(1000)
1000
>>> f.readline()

The filesystem on which "test-file" resides has a block size of 4,096 bytes, so on the backend, I would expect Python to read 3,096 bytes so the next read will be aligned with the filesystem blocks. What actually happens is that Python reads a 4,096 bytes. This is the output from an strace attached to the interpreter above:

Process 16543 attached
lseek(4, 0, SEEK_CUR)                   = 0
lseek(4, 1000, SEEK_SET)                = 1000
read(4, "\000\000\000\000\000\000\000\000\000\000"..., 4096) = 4096

----------
components: IO
messages: 246931
nosy: ericpruitt
priority: normal
severity: normal
status: open
title: Buffered I/O does not take file position into account when reading blocks
type: behavior
versions: Python 2.7, Python 3.2, Python 3.3, Python 3.4

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue24666>
_______________________________________