7

I have a script that I run as part of my development (it's the activation script for a Python venv environment). The way the documentation suggests that we run it is by going to the folder containing our venv folder and running . venv/bin/activate. This works properly (second command in my example).

However, what I would normally have run is ./venv/bin/activate, i.e. providing a relative path to the activate script (first command in my example). This doesn't work at all (although I am not surprised, because the activate file doesn't have "execute" permissions attached to it).

My-MBP:flask-tutorial stephendewey$ ./venv/bin/activate
-bash: ./venv/bin/activate: Permission denied
My-MBP:flask-tutorial stephendewey$ . venv/bin/activate
(venv) My-MBP:flask-tutorial stephendewey$

What is the command that works (. venv/bin/activate) actually doing? I've never seen syntax like that before.

  • Note to editors. The author is dealing with a script that is used in python development. But the question and definitely the answers are not python specific – Mark Mar 1 at 19:31
8

The . command is an alias for source so the two commands are really

./venv/bin/activate

and

source venv/bin/activate

Also note that for the system to actually process a file it needs the absolute path ie one beginning with /

Both files names here are relative ones ie ones that can only be understood with the knowledge of the current directory which is the variable $(CWD)

The two file names expand to $(CWD)/./venv/bin/activate and $(CWD)/venv/bin/activate . is the current directory and so both are $(CWD)/venv/bin/activate

The difference between running a command directly or via source is that if run directly as in the first command bash creates a new sub shell and runs the command in that and the commands in the script only affect that sub shell and when the script rends that sub shell is closed and all the changes to environment are lost. source, however, runs the command in the current shell and any changes to the environment remain after the script finishes as if the commands in the script were typed into the current shell.

The activate script (I assume is from Python virtual environment management) works by changing the $PATH so that the correct python environment is found when you use python script.py To do this you need to alter your current $PATH and so the activate script needs to be run using source.

Also see https://askubuntu.com/questions/182012/is-there-a-difference-between-and-source-in-bash-after-all and https://superuser.com/questions/176783/what-is-the-difference-between-executing-a-bash-script-vs-sourcing-it

Not that running a command requires the command to be executable. The command to be run is always a file and that file has to be marked as executable by the system. That is the executable flag has to be set on the file permissions which you can see by ls -l venv/bin/activate

source however is in the current shell and just reads the file as text and then executes the commands it sees, So this file only needs to be readable. For more on that see https://unix.stackexchange.com/questions/291404/why-does-bashs-source-not-need-the-execution-bit I like this answer

It's more of a convenience thing: Let the system run it for me directly if the the bit is set, otherwise I need to do it indirectly

  • 1
    Thanks, this answers almost all my questions. But why would the "direct command" run into a permissions issue while the "source command" doesn't? – Stephen Mar 1 at 17:14
  • @Stephen When you do a long listing, does the file lack the execute permission? ls -l venv/bin/activate portion of the answer? That would explain the permission error when you try to execute the file directly as opposed to letting the shell read the file for contents. Unix permissions come from a model established quite a while back, so they sometimes are confusing why you could read a file as opposed to execute it as a "binary" or "app" +1 to the asker and answerer here - great question that lets Mark explain why it happens as well as what to do. – bmike Mar 1 at 18:19
  • Thanks all, I think the edit to the answer answers my remaining question (the execute bit was indeed turned off, but I don't need to change that since I can just source it). – Stephen Mar 1 at 18:57
  • Important to note that running an executable and source-ing it are more different than that. Running the executable calls the system's exec function, whereas calling source on a file will read the file and run the lines of the file similar to if they were directly typed into the terminal. This is why you can run a binary program, but you can't source it. (eg. source /bin/ls vs /bin/ls) – Kevin Johnson Mar 2 at 0:31
  • Strictly speaking, source is an alias for .. The . command was the original. – JdeBP Mar 2 at 9:50
7

From man bash:

. filename [arguments]
source filename [arguments]
    Read and execute commands from filename in the current shell environment and
    return the exit status of the last command executed from filename. If filename
    does not contain a slash, filenames in PATH are used to find the directory
    containing filename. The file searched for in PATH need not be executable.

So basically source runs the code within the script/file as part of the current shell environment (which is different from executing it, which runs it in a separate shell). This is mainly used to set environment variables, aliases etc to be used in the current shell.

  • Thanks, I think the last part of the bash manual entry answers the part of my question about permissions: source doesn't even require something to be executable. – Stephen Mar 1 at 17:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .