Tom Mueller wrote:
> Shawn,
> Thank you for the comments.
> Shawn Walker wrote:
>>
>> history.py:
>>   line 555: a tuple instead of a list?
> My Python newness is showing.  Either a list or a tuple works here.  Why 
> is a tuple preferable to a list?

Minor efficiency; lists have more overhead than tuples.  So if it's just 
a list of values to iterate over, there's no reason to use a list.

>> cli/t_actuators.py:
>>   lines 167, 172: let's not use /bin/sh.  who knows what shell that 
>> will be?  instead, specify /usr/bin/bash or /usr/bin/ksh or 
>> /usr/bin/ksh93.  /bin/sh is undefined.
> None of those shells exist on Linux either. Linux doesn't include any 
> shells in /usr/bin. I had thought that /bin/sh was a pretty standard way 
> to reference a POSIX shell.

Hrm, actually, you're right.  My recollection of this was wrong.

If you have a script that needs anything beyond POSIX shell 
functionality it should be /bin/bash, etc.  But if you're only using 
POSIX shell functionality, #!/bin/sh is the right thing to do [1].

>>
>> cli/testutils.py:
>>   line 35: given that you trap this import as you did pspawn, why not 
>> check for the presence of pwd in globals as you did for pspawn first?
> Because get_su_wrap_user is never called on systems that don't have pwd 
> (Windows). On Windows we only run the API tests.  But the test framework 
> reads in the Python modules (including testutils.py) for all tests even 
> though they are not run. To get the CLI tests to run on Windows is a 
> much bigger task that I'm not taking on this this changeset.
> 
> Would it be better to localize the import of pwd to the get_su_wrap_user 
> method?

Danek would be a better person to answer that, I'm fine with your 
explanation.

Cheers,
-- 
Shawn Walker

[1] 
http://mail.opensolaris.org/pipermail/indiana-discuss/2007-September/002165.html
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to