Fredrik Lundh wrote: > Brian Mills wrote: > > > There's another (IMHO more readable) way to do it if you can afford > > defining a short little "compare" function, and telling <list>.sort() > > to use that instead of its default: > > > >>>> def myListCmp(lst1, lst2): > > ... if lst1[0] < lst2[0]: return -1 > > ... if lst2[0] > lst2[0]: return 1 > > ... return 0 > > shorter: > > def myListCmp(a, b): > return cmp(a[0], b[0])
Good point. > but using a compare function instead of a key mapper is not good advice, > in general. brief discussion here: > http://effbot.org/pyfaq/i-want-to-do-a-complicated-sort-can-you-do-a-schwartzian-transform-in-python Is this mostly because of the stability problem described here: http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/52234 ? Or is it more a performance issue due having to make so many function calls? > > also note the OP didn't specify what to do for records where the first > column was identical, so I guess a plain sort() call, without any custom > compares or mappings, would work as well as the fancier alternatives... I can't believe I didn't try this, but for the given example it works perfectly. It even acts gracefully when you give it such sociopathic lists as >>> c = [[3, 1], [8, 2], [6, 3], [12, 4], [1, 5], ["oh noes!", 6], [24, 7], [ope n("file.txt", 'r'), 8], [[1, 2, 3], 9]] >>> c.sort() >>> c [[1, 5], [3, 1], [6, 3], [8, 2], [12, 4], [24, 7], [<open file 'file.txt', mode 'r' at 0x00A65608>, 8], [[1, 2, 3], 9], ['oh noes!', 6]] Though changing the ordering system it decides to use may take some more research. -- http://mail.python.org/mailman/listinfo/python-list
