PyCon 2010: The Mighty Dictionary

PyCon 2010: The Mighty Dictionary
Presented by Brandon Craig Rhodes

[@k3v1l]

May your hashes be unique, Your Has tables never full. And may your keys rarely collide as well

[@RohanAnand417]

Amen!

[@ComputerVisionPro]

Ameen

[@akshy471]

After watching this, I bet I'll land that CEO role in Google.

[@DenisBTV]

The clearest Python dictionary background explanation! Thanks!

[@Naz-yi9bs]

The way he says "Baaaaal" dying. Amazing detailed presentation, thank you!

[@kartikeysingh9764]

"Equal values should have equal hashes regardless of their types" one of the most important points I found from this talk and also, copy() the dict to shrink it down if lots of deletion.

[@surashreekulkarni6115]

Kickass!! Really well explained!

[@drizzyDreF]

This was very helpful

[@tomaszszulc7259]

Awesome talk!

[@aaronhall8039]

Where can we find specific documentation about this, or will we have to derive an understanding of this by using a C debugger or a static analysis of the C source?

[@sohangchopra6478]

why is the video quality so bad???

[@vikingden8402]

The clearest Python dictionary background explanation! Thanks! BTW, slides: rhodesmill.org/brandon/slides/2010-03-pycon/

[@hamman_samuel]

Used words from an English dictionary to input elements into a Python dictionary. And all the undergrads screamed in agony.

[@hiwaqas]

When going for lookup, how is the path to a certain key known? Is it stored/cached when the dict was being formed? Or is the path taken, dependent on the higher bits of the hash?

[@justgivemethetruth954]

24:26

[@vincentphilippe_18]

A hash function return the same hash for one value. So given any value, a hash function can retrieve where a data is stored. That's why here it use bit() function because every value have a unique 'bits' value representation.

[@justgivemethetruth954]

So, does the lookup then try to match the key every time the has points to a given slot? And then if the key goes not match, it goes through successive keys until it finds a match? This would seem to say to me that unless you need to and cannot use a list, stay away from dicts? And so this is really a trick lecture ... only a problem with miniscule sizes of dicts, but seems like you always are wasting 1/3 of the space allocated for your dict addressiing since it will resize, quadrupling or doubling when it gets to 2/3? No problem, RAM is cheap!

[@4rph3n]

What is the "bits" function you're using?

[@aaronhall8039]

+Sebastian Wozny It is something like: def bits(integer): return bin(integer)[2:]

[@jursamaj]

It's right there in the video at 0:12: def bits(n): n+=2**32 return bin(n)[-32:]

[@dzidmail]

What if value is a large string that doesn't fit between two hashed memory pointers? I think something is missing from how this really works.