On 17/03/2026 12:08, Janis Papanagnou wrote:If there are simple better algorithms there's no reason butOn 2026-03-17 12:47, Bart wrote:No, timing is considered relative to the rest of the task.On 17/03/2026 04:38, Janis Papanagnou wrote:>On 2026-03-17 01:29, Bart wrote:>>>
(Also I quite like using bubble sort because so many deride it!)
How miserable! (I feel so sorry for you.)
Why would that be miserable?
Because of the reason you pretend for using it.
>[...]>
>
Bubble-sorting even 1000 strings takes about 10ms, which is small fraction of overall build-time of a library exporting 1000 functions.
>
For a mere 100 functions, sort time is negligible.
And because of reason here based on arbitrary absolute times instead
of actual algorithmic complexity.
If this ever became a bottleneck then it is a trivial upgrade to a better routine.
ignorance to not use them in the first place! - It's really
stupid to deliberately use inferior, or (as here) even the
worst existing algorithms. - What do professionals? - Let's
have a look at a Quicksort implementation on a CDC mainframe
back in the 1980's. They implemented it in a way that scales
according to the CS state-of-the-art; divide-et-impera for
the Quicksort, pick pivot element from a set of three, and
sort divided runs of length smaller than 10 with a straight
insertion sort algorithm. - No educated computer scientist
would have decided to choose Bubblesort in any place here.
Let alone for your "reason" that this algorithm is "derided".
There's a reason why it's derided; that's called scientific
analysis of its quality.
Janis
[...]
Les messages affichés proviennent d'usenet.