| 1 |
On Wed, 11 Jun 2008 04:15:35 -0700 |
| 2 |
Brian Harring <ferringb@×××××.com> wrote: |
| 3 |
> Being that you can't understand the problem you're commenting on, |
| 4 |
> I'll explain it for you. |
| 5 |
> |
| 6 |
> While you can remove _p1, or _<random_suffix> you cannot change the |
| 7 |
> ordering of an existing version component. Simple example you should |
| 8 |
> grok, changing of 1_p1 such that it's <1.0 is not valid. |
| 9 |
> |
| 10 |
> As I've indicated repeatedly in this thread, and y'all have missed, |
| 11 |
> you cannot change the semantics of the ordering. Sure, you could |
| 12 |
> remove a version component from usage- that said, you cannot change |
| 13 |
> it's ordering. |
| 14 |
|
| 15 |
Except that isn't what you said. What you said is: |
| 16 |
|
| 17 |
> One thing I'll note is that the .ebuild-$EAPI approach isn't the end |
| 18 |
> all fix to versioning extensions that y'all represent it as. |
| 19 |
> Essentially, what .ebuild-$EAPI allows is additions to version |
| 20 |
> comparison rules, no subtractions. Each new $EAPI *must* be a |
| 21 |
> superset of previous $EAPIs. |
| 22 |
|
| 23 |
Which is obviously untrue. Here's a perfectly consistent, perfectly |
| 24 |
implementable (but probably not desirable) rule for versions for a |
| 25 |
future EAPI that is not a strict subset, nor is it additions: |
| 26 |
|
| 27 |
"Any package using EAPI 2 may not use any version component that uses |
| 28 |
leading zeros. Any package manager supporting EAPI 2 must support scm." |
| 29 |
|
| 30 |
> Further, you cannot change the ordering of an existing version- if |
| 31 |
> you can't understand why shifting 0.006 to equivalent to 0.6, then |
| 32 |
> frankly, this discussion need not continue. |
| 33 |
|
| 34 |
http://en.wikipedia.org/wiki/Straw_man |
| 35 |
|
| 36 |
-- |
| 37 |
Ciaran McCreesh |
Archives