[IPython-user] Killing a hung program

Fernando Perez Fernando.Perez at colorado.edu
Fri Jun 17 11:34:18 CDT 2005


Prabhu Ramachandran wrote:
>>>>>>"Fernando" == Fernando Perez <Fernando.Perez at colorado.edu> writes:
> 
> 
>     Fernando> You can also keep a log of all commands you run via
>     Fernando> %logstart.  The docstrings for the logging-related
>     Fernando> functions have details.
> 
> A couple of suggestions (apologies if they are dumb or already
> implemeted :).  logstart sticks a logfile in each directory you start
> IPython from.  I thought it would be cool to have the following
> options:
>  
>   1. Allow the user to have one logfile that always logs stuff into
>      .ipython/log.  Unlike __IP.savehist, this log is always
>      up-to-date.

Well, what do you do there with a user who is running (as I often do) an 
ipython session in each of 5 concurrently open shells?  They are going 
to trample each other pretty badly, no?

> 
>   2. An option to specify the log mode from the command line.  Right
>      now you can say ipython -l, but that by default uses the rotate
>      mode (which I dislike).  I ran into this problem sometime back
>      but did not get the chance to post about it.

Yes, we could add a new command-line switch for this, though I fear that 
ipython already has too many options at the CL.  What about just having 
in your rc file a call to %logstart in any way you like?  That's the 
beauty of the rc file: you can set your ipython session to behave 
_exactly_ the way you personally want it to, with arbitrarily complex 
configurations, without cluttering the public interface with too many 
details.

If anything, in the future (as the rc files become regular python code) 
I'm inclined to have _less_ command-line options rather than more, and 
leave all complex configuration to be done in the rc file.

Any thoughts?

Cheers,

f




More information about the IPython-user mailing list