1 |
On Mon, 2004-09-06 at 19:16, Ned Ludd wrote: |
2 |
> Do you really have to create another category? |
3 |
|
4 |
I wrote a little script to help in this decision. It suggests a range |
5 |
for category sizes, based on the average size plus or minus a multiple |
6 |
of the standard deviation (SD). The default multiple is 1/2 the SD. |
7 |
|
8 |
The conclusions are: If you have a category larger than the suggested |
9 |
max, split it. If it's smaller than the suggested min, don't create it. |
10 |
|
11 |
Perhaps the SD multiple is worth arguing over, but 1/2 seemed |
12 |
reasonable, based on the results. |
13 |
|
14 |
Another thing to consider is that one of the goals of categories is to |
15 |
make the tree browseable. Therefore we should keep the number of things |
16 |
to be browsed roughly equal at every level. So the number of categories |
17 |
should be roughly equal to the average number of packages per category. |
18 |
|
19 |
Anyway, there's a couple of things to think about. |
20 |
-- |
21 |
Donnie Berkholz |
22 |
Gentoo Linux |