When you said "C-runtime buffered I/O", are you talking about Standard I/O in C (FILE * object)? AFAIN, In CPython, we didn't use Standard I/O, right?
Dennis Lee Bieber <wlfr...@ix.netcom.com> 于2019年6月25日周二 上午12:48写道： > On Mon, 24 Jun 2019 15:18:26 +1200, DL Neil > <pythonl...@danceswithmice.info> declaimed the following: > > > > > >However, the OpSys may have read considerably more data, depending upon > >the device(s) involved, the application, etc; eg if we ask for 2 bytes > >the operating system will read a much larger block (or applicable unit) > >of data from a disk drive. > > > > Depending upon implementation, there could be layers of buffers > involved... > > Python application requests, say, 50 bytes using a "buffered I/O" > file > object. That file object may invoke a C-runtime buffered I/O call that > requests whatever the C-runtime buffer-size is -- say a 512 byte sector. > That request goes to a low-level device driver for a file system/device > that does I/O in 4KB clusters. So the first read results in the OS reading > 4KB into a buffer, and passing 512 bytes to the C-call, which then returns > 50 bytes to the Python layer. > > The second read for 50 bytes is satisfied from the remaining bytes > in > the C-runtime sector buffer. The 11th read of 50 bytes will get 12 bytes > from the sector, and then the C-runtime has to request the next sector from > the OS, which is satisfied from the file system cluster buffer. After the > 8th sector is processed, the next request results in the OS going to disk > for the next cluster. > > > -- > Wulfraed Dennis Lee Bieber AF6VN > wlfr...@ix.netcom.com > http://wlfraed.microdiversity.freeddns.org/ > > -- > https://mail.python.org/mailman/listinfo/python-list > -- https://mail.python.org/mailman/listinfo/python-list
