How does Linux differentiate between opening a file in read mode and actually reading data from it?

Question

I created a simple text file and checked its timestamps using stat and here is the output:

stat Output before any python script

papagolf@Sierra ~/My Files/My Programs/Learning/Python % stat prac09_timestamp.txt 
  File: prac09_timestamp.txt
  Size: 10          Blocks: 8          IO Block: 4096   regular file
Device: 802h/2050d  Inode: 2485542     Links: 1
Access: (0600/-rw-------)  Uid: ( 1000/papagolf)   Gid: ( 1001/papagolf)
Access: 2017-09-10 11:10:40.979845492 +0530
Modify: 2017-09-10 11:10:40.979845492 +0530
Change: 2017-09-10 11:10:40.979845492 +0530
 Birth: -

Then I ran a python script that simply opens the file in read mode. I was expecting the time "Access" stamp to change but it didn't.

Python Script and output of stat after running it

papagolf@Sierra ~/My Files/My Programs/Learning/Python % cat python prac09_timestamp.py 
cat: python: No such file or directory
a = open("prac09_timestamp.txt", "r")
a.close()
papagolf@Sierra ~/My Files/My Programs/Learning/Python % python prac09_timestamp.py 
papagolf@Sierra ~/My Files/My Programs/Learning/Python % stat prac09_timestamp.txt 
  File: prac09_timestamp.txt
  Size: 10          Blocks: 8          IO Block: 4096   regular file
Device: 802h/2050d  Inode: 2485542     Links: 1
Access: (0600/-rw-------)  Uid: ( 1000/papagolf)   Gid: ( 1001/papagolf)
Access: 2017-09-10 11:10:40.979845492 +0530
Modify: 2017-09-10 11:10:40.979845492 +0530
Change: 2017-09-10 11:10:40.979845492 +0530
 Birth: -

But when I change the script to actually read the data, the time stamp changed. Can you explain it?

New Python Script and output of stat after running it

papagolf@Sierra ~/My Files/My Programs/Learning/Python % cat prac09_timestamp.py 
a = open("prac09_timestamp.txt", "r")
a.readlines()
a.close()
papagolf@Sierra ~/My Files/My Programs/Learning/Python % python prac09_timestamp.py
papagolf@Sierra ~/My Files/My Programs/Learning/Python % stat prac09_timestamp.txt 
  File: prac09_timestamp.txt
  Size: 10          Blocks: 8          IO Block: 4096   regular file
Device: 802h/2050d  Inode: 2485542     Links: 1
Access: (0600/-rw-------)  Uid: ( 1000/papagolf)   Gid: ( 1001/papagolf)
Access: 2017-09-10 11:13:12.699848321 +0530
Modify: 2017-09-10 11:10:40.979845492 +0530
Change: 2017-09-10 11:10:40.979845492 +0530
 Birth: -

Answer 1

I don't think this is a Python question at all. It's system level. And at that level opening a file is an entirely distinct operation from reading a file. I'm not an expert by any stretch of the imagination on Linux internals but here are some basic facts:

• A file is opened with the open(2) system call.
• A file is read with the read(2) system call
• There is a flag for open(2) that instructs the system to not update access time when read(2) is called. It's called O_NOATIME
• You can infer from this that open(2) doesn't update access time

If you think about what the word access means this makes sense. open(2) returns a file descriptor that can subsequently be used to read the file. To use a crude analogy, it's like having an all-access pass to a concert. Until you show up and start hitting on groupies back stage you haven't accessed anything. ;)