Using python to start programs after logging in

[MRAB]

On 2017-01-19 23:36, Cecil Westerhof wrote:
> On Thursday 19 Jan 2017 22:21 CET, Cecil Westerhof wrote:
>
>> On Thursday 19 Jan 2017 21:12 CET, MRAB wrote:
>>
>>> On 2017-01-19 19:08, Cecil Westerhof wrote:
>>>> I am writing a python program to start the programs that need to
>>>> be started after logging in.
>>>>
>>>> I have the following imports:
>>>> from subprocess import check_call, Popen, STDOUT
>>>> from time       import sleep, strftime
>>>>
>>>> And use the following code: check_call(tuple('wmctrl -s
>>>> 10'.split())) log_file =
>>>> open('Logging/firefox_%T.log'.replace('%T', strftime('%F_%R')),
>>>> 'w') Popen(tuple('firefox'.split()), stdout = log_file, stderr =
>>>> STDOUT)
>>>>
>>>> The first statement is to go to the correct desktop.
>>>>
>>>> Is this a good way to do things, or could I do it in a better way?
>>>>
>>> Apart from the other answer, your use of .replace is odd. This
>>> would be better:
>>>
>>> log_file = open('Logging/firefox_%s.log' % strftime('%F_%R'), 'w')
>>
>> The reason is that it is ‘generic’ code. In the future instead of
>> string it will be a variable and only when the variable contains %T
>> it should be replaced.
>>
>>
>>> Even better would be to use the 'with' statement as well.
>>
>> The same here. Not always has the output to be logged and then I
>> would use NONE for log_file.
>
> I wrote a function for switching to the correct virtual desktop and
> starting all the commands. I am also using with now:
> def do_desktop(desktop, commands, seconds_to_wait = 10):
>     desktop_command = ('wmctrl -s ' + desktop).split()
>     check_call(tuple(desktop_command))

You don't have to pass a tuple to check_call; it'll also accept a list.

>     for command_arr in commands:
>         command         = command_arr[0].split()
>         log_directory   = command_arr[1]
>         directory       = command_arr[2]
>         if (directory != ''):
>             chdir(directory)

There's no need for the parentheses around the 'if' condition.

>         if (log_directory == 'NONE'):
>             Popen(tuple(command.split()))
>         else:
>             log_file_name = log_directory.replace('%T', strftime('%F_%R'))
>             with open(log_file_name, 'w') as log_file:
>                 Popen(tuple(command), stdout = log_file, stderr = STDOUT)
>         if (directory != ''):
>             set_default_dir()

Using 'chdir' is generally a bad idea.

'Popen' will start the process and then continue, so what you're doing 
is changing the directory, starting a process, and then changing the 
directory again, possibly before the process has truly started (i.e. 
you've told the OS to start it, but there might be a small delay before 
that happens). Popen accepts a 'cwd' argument, which sets the 'current 
working directory' for that process.

>     sleep(seconds_to_wait)
>
>
> The function set_default_dir:
> def set_default_dir():
>     chdir(expanduser('~'))
>
>
> The example with firefox is then done with:
> commands = [
>     ['firefox', 'Logging/firefox_%T.log', ''],
> ]
> do_desktop('10', commands, 30)
>
>
> Sometimes a command needs to be executed in a different directory,
> that is done like:
> commands = [
>     ['lein run',                                                                    'NONE', 'Clojure/Quotes'],
>     ['xfce4-terminal --maximize --execute screen -S Clojure -c ~/.screenrcClojure', 'NONE', ''],
> ]
> do_desktop('6', commands, 30)
>
Why 'NONE'? Why not None?