| 1 |
On Fri, Jun 13, 2008 at 10:43:39AM +0200, Luca Barbato wrote: |
| 2 |
> Anyway pkgcore and portage devs, I'd like your opinion on this point. |
| 3 |
|
| 4 |
Custom repository is how I intended to implement this; the upshot of |
| 5 |
the version translation is that the resultant vdb version is managable |
| 6 |
by any PM, regardless if they support -live, which 'generated' the |
| 7 |
ebuild. |
| 8 |
|
| 9 |
Presuming svn python bindings aren't as sucky as I recall, tempted to |
| 10 |
take a thwack at it over the weekend. |
| 11 |
|
| 12 |
One thing that would need ironing out for such a proposal is how to |
| 13 |
mark in the template metadata variation dependant upon vcs exceeding a |
| 14 |
certain level- example would be, |
| 15 |
|
| 16 |
< revno 300 |
| 17 |
RDEPEND='=dev-lang/python-2.4*' |
| 18 |
|
| 19 |
>= revno 300 |
| 20 |
RDEPEND='>=dev-lang/python-2.5' |
| 21 |
|
| 22 |
That... gets tricky. I can think of hacks for it, but none I like. |
| 23 |
Alternatively, if there was a way to have seperate templates that are |
| 24 |
selected dependant upon the desired rev, it becomes possible. |
| 25 |
|
| 26 |
One other thing that needs discussion imo, is how such a scheme would |
| 27 |
work for non integer based revnos- git for example, which is reliant |
| 28 |
on a hash (just the hash, afaik). |
| 29 |
|
| 30 |
Either way, I tend to view it as having a fair bit more control then |
| 31 |
the -scm proposal, although needs fleshing out a bit. |
| 32 |
|
| 33 |
~harring |
Archives