| 1 |
On Fri, 17 Dec 2010 17:27:05 +0100 |
| 2 |
Sebastian Luther <SebastianLuther@×××.de> wrote: |
| 3 |
> Am 17.12.2010 16:25, schrieb Ciaran McCreesh: |
| 4 |
> > So would anyone be especially opposed to making "best leftmost" an |
| 5 |
> > explicit requirement, enforced by repoman where possible (at least |
| 6 |
> > for the >= / < case)? |
| 7 |
> |
| 8 |
> Why can't the PM handle >= / < cases itself? |
| 9 |
|
| 10 |
Because things are almost never as simple as 'just' >= / <. You can add |
| 11 |
in clever trickery to deal with very specific cases, but the second |
| 12 |
someone throws things off by adding in a use dependency or a third |
| 13 |
package, things get weird. |
| 14 |
|
| 15 |
Consider a variation on the original case: || ( <a-2 >=a-2[x] ) where |
| 16 |
the user has specified -x for a. What should happen then? |
| 17 |
|
| 18 |
What about || ( <a-2[x] b >=a-2 ) ? Should that be rewritten in the same |
| 19 |
way? |
| 20 |
|
| 21 |
What about || ( <a-2[x] ( >=a-2 b ) ) ? Should the package mangler be |
| 22 |
clever enough to figure that one out too? What if b isn't already |
| 23 |
installed there? |
| 24 |
|
| 25 |
Which is really the problem: clever heuristics get extremely |
| 26 |
complicated very quickly, and they're never enough. |
| 27 |
|
| 28 |
-- |
| 29 |
Ciaran McCreesh |
Archives