Preallocate? -- potentially brain dead question about performance

[Jan Danielsson]

Hello all,

I have a list which contains a bunch of tuples:

mylist = [ ('1', 'Foobar'), ('32', 'Baz'), ('4', 'Snorklings') ]

(The list can potentially be shorter, or much longer). Now I want to
take the first element in each tuple and store it in a list (to use in a
Set later on).

Which would You prefer of the following:

newlist = [ ]
for e in mylist:
newlist.append(int(e[0]))

..or..

newlist = [ None ] * len(mylist)
for i in range(len(mylist)):
newlist.append(int(e[0]))


To me, the second one is more appealing, when I think in terms of
risk of memory fragmentation, etc. But since Python is such a high level
language, I'm not sure my traditional reasoning applies.

 

[bearophileHUGS]

Jan Danielsson:

newlist = [ None ] * len(mylist)
for i in range(len(mylist)):
newlist.append(int(e[0]))

Note that this appends after the Nones, not over them. And note that
here the name 'e' is undefined.

The most used idiom for that is:

newlist = [int(e[0]) for e in mylist]

Or better lazily if you want to create a set, avoiding the list
creation:

newlist = set(int(e[0]) for e in mylist)

Bye,
bearophile

 

[Jan Danielsson]

newlist = [ None ] * len(mylist)
for i in range(len(mylist)):
newlist.append(int(e[0]))

Note that this appends after the Nones, not over them. And note that
here the name 'e' is undefined.

What an idiot I am.. Yes, I know that.. And I actually *had* working
code laying around, but rather than opening it and copy-and-pasting, I
simply rewrote it. I obviously meant:

newlist = int(e[0])

The most used idiom for that is:

newlist = [int(e[0]) for e in mylist]

Or better lazily if you want to create a set, avoiding the list
creation:

newlist = set(int(e[0]) for e in mylist)

...completely avoiding the design issue I raised altogether. Thanks!
Exactly what I was hoping for!

 

[bearophileHUGS]

Jan Danielsson:

...completely avoiding the design issue I raised altogether. Thanks!
Exactly what I was hoping for!

Another common way to do it, it may be a little slower because that
n,txt is a tuple:

items = set(int(n) for n,txt in mylist)

Bye,
bearophile