| 1 |
On 05/07/2015 01:06 AM, Alexander Berntsen wrote: |
| 2 |
> On 24/04/15 18:32, Zac Medico wrote: |
| 3 |
>> - installed_package = installed_package[-1] |
| 4 |
>> + try: |
| 5 |
>> + self._vardb.match_unordered |
| 6 |
>> + except AttributeError: |
| 7 |
>> + installed_package = installed_package[-1] |
| 8 |
>> + else: |
| 9 |
>> + installed_package = portage.best(installed_package) |
| 10 |
>> + |
| 11 |
> This probably makes sense, but I don't understand how this helps to |
| 12 |
> alleviate the problem. Would you mind explaining it? |
| 13 |
|
| 14 |
This hunk just fixes to ensure that it shows the highest version in the |
| 15 |
case were there are multiple slot installed. Otherwise, since |
| 16 |
IndexedVardb returns unordered match results, it could randomly select a |
| 17 |
lower version from another slot. |
| 18 |
|
| 19 |
> The rest of the patch looks OK, even if I dislike the strong imperative |
| 20 |
> nature of the replaced versions removal. |
| 21 |
|
| 22 |
I'm open to suggestions if you can think of a nicer way to express it. |
| 23 |
It certainly be nicer if the data from vdb_metadata.pickle had the |
| 24 |
packages indexed by ${CATEGORY}/${PN}, so that the sequential search |
| 25 |
would not be necessary. We should do that later if we decide to modify |
| 26 |
the vdb_metadata.pickle format in a backward-incompatible way. |
| 27 |
-- |
| 28 |
Thanks, |
| 29 |
Zac |
Archives