| 1 |
Hello, |
| 2 |
|
| 3 |
(Time for some more growing pains. It is my opinion that prefix keywords |
| 4 |
are one of the last major blockers left in our effort to merge back to |
| 5 |
gentoo-x86. Inspired by chatting with solar in irc) |
| 6 |
|
| 7 |
The current situation is when a package is imported, it automatically |
| 8 |
gets a few keywords that gentoo-x86 has. The dev that imports it, tests |
| 9 |
that it compiles and leaves the 'automatic' keywords. Version bumps |
| 10 |
happen automatically and the keywords are carried forward without |
| 11 |
testing. If there is an issue we react to it. Besides the initial |
| 12 |
importing (which may have happened years ago) we don't do any proactive |
| 13 |
testing[1]. |
| 14 |
|
| 15 |
So, since we are already in a hugely reactive mode..why don't we just |
| 16 |
get rid of prefix keywords completely? It gets hairy if the arch most |
| 17 |
always needs patches (FreeMiNT/IRIX comes to mind). However, this is one |
| 18 |
reason that we ask for everyone's help in submitting patches upstream. |
| 19 |
|
| 20 |
Before anyone says "but, that will be much more likely to break my |
| 21 |
prefix" - I refute that because we are already running on this policy |
| 22 |
with regards to the automatic bumps. For the most part, it is smooth. |
| 23 |
Major packages are masked if someone hasn't tested them yet (eg. gcc & bash) |
| 24 |
|
| 25 |
Any other thoughts? |
| 26 |
|
| 27 |
-Jeremy |
| 28 |
|
| 29 |
[1]: I do weekly emerge -e world but that clearly will not test all |
| 30 |
packages on all arches. Just the @system+$random set on one arch. |
Archives