| 1 |
On Sat, Oct 22, 2005 at 12:13:42PM +0900, Jason Stubbs wrote: |
| 2 |
> Something like: |
| 3 |
> * Add base class(es) for new cache framework |
| 4 |
> * Add cache backend for XYZ database |
| 5 |
> * Switch portdbapi to the new framework |
| 6 |
> * Remove old framework |
| 7 |
eclass_cache.py chunking (portage.py removal) |
| 8 |
cache replacement (base + implementations) |
| 9 |
portage.py (dbapi), emerge changes (integration of new cache). |
| 10 |
removal patch |
| 11 |
|
| 12 |
That said... would be curious about suggestions on how to do this |
| 13 |
sanely. Chunking the beast up (patch jockeying) after the fact I can |
| 14 |
do, but in instances like this... it's not easy to chunk it down into |
| 15 |
features/tweaks. Basically is big ass blobs of "new stuff", |
| 16 |
"conversion to new stuff", "remove old stuff". |
| 17 |
|
| 18 |
Even with that... still is tricky. |
| 19 |
|
| 20 |
Offhand, the existing cache patch could be reduced pretty heavily by |
| 21 |
breaking it down into addition, and removal of old cache. |
| 22 |
|
| 23 |
Obviously after review ability, but also would note the further down |
| 24 |
you push the required granularity of the patches, the harder it is to |
| 25 |
have a big picture view of the patchset changes (imo). |
| 26 |
|
| 27 |
|
| 28 |
> > So... guideliness. ? |
| 29 |
> |
| 30 |
> Should start a new thread about it later. I'd like to get this one finalized |
| 31 |
> first. :) |
| 32 |
Personally, I'm kind of inclined to just have people state stuff. |
| 33 |
a commit message of "fixed shit" doesn't really cover it, obviously, |
| 34 |
but I'd say what's been coming in on the portage-commits alias as of |
| 35 |
late covers it (both 3.0 and 2.0 commit messages) |
| 36 |
|
| 37 |
Seperate discussion maybe, but kind of think of it as a no-brainer :) |
| 38 |
~harring |
Archives