New restrain builtin function?

[Pierre Rouleau]

John Roth wrote:

> "Pierre Rouleau" <prouleau at impathnetworks.com> wrote in message
> news:xK17c.26714$Eb6.852981 at news20.bellglobal.com...
> 
>>Hi all,
>>
>>In several occasions, I found myself looking for a function that would
>>take a value and restrict is within a specified set of boundaries.   For
>>a 1-dimension value, I could simply write
>>
>>min(max(value,theMin),theMax)
>>
>>to restrict value within the range made of theMin to theMax.
>>It assumes that theMax >= theMin.
>>
>>I was looking for a single call to restrict a value within bounds that
>>would do pretty much that, so i wrote this trivial one:
>>
>>def restrain(value, theMin, theMax) :
>>     """Return a value that is in restricted to the [theMin, theMax]
> 
> range.
> 
>>     **Example**
>>
>>     >>> for val in xrange(-1,7,1):
>>     ...   print "restrain(%d,1,5) = %d" % (val, restrain(val,1,5))
>>     ...
>>     restrain(-1,1,5) = 1
>>     restrain(0,1,5) = 1
>>     restrain(1,1,5) = 1
>>     restrain(2,1,5) = 2
>>     restrain(3,1,5) = 3
>>     restrain(4,1,5) = 4
>>     restrain(5,1,5) = 5
>>     restrain(6,1,5) = 5
>>     >>>
>>     """
>>
>>     assert(theMax >= theMin)
>>     return min(max(value,theMin),theMax)
>>
>>
>>Without the assertion check, the restrain() named function runs
>>expectedly slower than the explicit min(max()) calls:
>>
>>C:\Python23\Lib>timeit "min(max(5,1),0)"
>>1000000 loops, best of 3: 1.02 usec per loop
>>
>>C:\Python23\Lib>timeit -s"from ut import restrain" "restrain(5,1,6)"
>>1000000 loops, best of 3: 1.53 usec per loop
>>
>>Again, I found myself writing code that has to restrain the values to
>>some range and prefer using the restrain() function instead of the
>>min(max()) one.
>>
>>Therefore, I was wondering if it would it make sense to add a function
>>like restrain() to the list of Python built-ins.  Or is there something
>>like that already in the Python library?
>>
>>
>>Thanks
>>
>>Pierre
> 
> 
> I know I've wanted that on (infrequent) occasion. I think
> "minmax" makes a better name, though, and I suspect it
> would have to be implemented in C to make any sense,
> performancewise.
> 
> John Roth
> 
I would prefer 'limit' than 'minmax' myself if 'restrain' was not 
'retained' ;-)

Pierre Rouleau