| 1 |
On Tue, Nov 08, 2005 at 06:32:18PM +0100, Marius Mauch wrote: |
| 2 |
> On Tue, 8 Nov 2005 02:29:14 -0600 |
| 3 |
> > Question is how will it scale for non-bugfixes, disruptive changes |
| 4 |
> > like cache backport, elog backporting, confcache, etc? What I'm |
| 5 |
> > concerned about is what's going to occur with .5x when large changes |
| 6 |
> > start sliding into it (or into a minor)- basically the territory |
| 7 |
> > we're wandering into right now with cache/exec refactoring for .54. |
| 8 |
> |
| 9 |
> My 0.02 something: |
| 10 |
> Stick with the current mess for 2.0.x, if it turns out we really need |
| 11 |
> to push something out there are still options (like using a _p suffix |
| 12 |
> or a fourth component). Never make a 2.1.x release (this version is |
| 13 |
> "burned" already). |
| 14 |
|
| 15 |
Expanding to a 4th component is just duplicating a minor field. |
| 16 |
|
| 17 |
2.1 being out there is a fricking mess, but I'm not particularly game |
| 18 |
to continue bad practices just because the portage versioning scheme |
| 19 |
already has blown it's own foot off. |
| 20 |
|
| 21 |
|
| 22 |
> Why this even if it's a mess? People are used to it already, adding a |
| 23 |
> 2.x with x>0 could be interpreted as "the next major version" with |
| 24 |
> use-depends and stuff, 2.1 is burned as stated above and currently it |
| 25 |
> seems to work. |
| 26 |
|
| 27 |
Use deps were never promised in 2.1 offhand. 2.1 stuff is being |
| 28 |
backported; yes the 2.1 alpha that got shoved out there isn't ever |
| 29 |
going to go stable (was dead 3 months prior to the release), but that |
| 30 |
doesn't mean we can't just spread the features out as we go. |
| 31 |
|
| 32 |
|
| 33 |
> Savior can and will reopen this discussion anyway. |
| 34 |
|
| 35 |
Portage 3 will be sane versioning scheme, that's a given. |
| 36 |
However, that's not a good reason to ignore the current problem. |
| 37 |
|
| 38 |
We've already screwed with the versioning scheme to get away from |
| 39 |
the previous bad setup, I'd prefer to finish it now rather then |
| 40 |
wait N months and again screw with versioning. |
| 41 |
|
| 42 |
The point of version components is to convey some extra meaning- |
| 43 |
having people guessing at what it is doesn't help that in anyway, |
| 44 |
hence any change of the components *should* be done once, and stuck |
| 45 |
to. |
| 46 |
|
| 47 |
I'd posit we're still in that timeframe for ironing out what versioning |
| 48 |
scheme we'll use _longterm_, and implementing it now. |
| 49 |
~harring |
Archives