| 1 |
Hi! |
| 2 |
|
| 3 |
On Wed, Sep 06, 2006 at 10:21:14AM +0200, Jos? Gonz?lez G?mez wrote: |
| 4 |
> for, let's say, a year WITH NO SECURITY BACKPORTS on them. This would be |
| 5 |
|
| 6 |
Impossible. There many means for 'stable'. For me, stable mean no |
| 7 |
security holes and no critical bugs as absolute requirements and all other |
| 8 |
things discussed under this topic in last year as less important requirements. |
| 9 |
|
| 10 |
For now I see only two alternatives for 'no security holes & no critical bugs' |
| 11 |
distribution: |
| 12 |
1) Use something like stable Debian/whatever distribution, which was |
| 13 |
released some time ago, and has old enough software with good support: |
| 14 |
updates with (backported) security fixes and critical bug fixes |
| 15 |
(or workarounds). Weakness of this way is having sometimes too old |
| 16 |
software, which is unacceptable because your customers will need |
| 17 |
features of newer software and you'll have to install & support it |
| 18 |
manually. |
| 19 |
2) Believe in Gentoo ARCH=x86 _IS_ 'stable' in all means and spend a little |
| 20 |
more time for compiliing/updating/testing. As bonus you'll have fairly |
| 21 |
up-to-date system with more features and your customers will love it. :) |
| 22 |
|
| 23 |
I vote for regular updates just because it's much more simpler to detect |
| 24 |
broken software and rollback to previous version when you update 1-3 |
| 25 |
packages at once, than when you update overall system once per year. |
| 26 |
|
| 27 |
If you think possibility of your application failure because of upgrading |
| 28 |
Gentoo is something which much never happens, even with fairly small |
| 29 |
possibility, then you just thinking wrong way. There always possibility for |
| 30 |
such failure because of bug in your application or hardware failure which |
| 31 |
you can't prevent! Of course we must do everything we can to minimize |
| 32 |
possibility of such application failure, including software update reason... |
| 33 |
But I think using Gentoo x86 and regular updates is good enough way to |
| 34 |
reach this goal. Statistics say SOME ppl have troubles in this setup in |
| 35 |
average once per year. I think it's good enough, and I hope Gentoo devs |
| 36 |
working on improving this statistics. :) And I'm sure same troubles happens |
| 37 |
even with stable Debian/whatever distribution from time to time. |
| 38 |
And will happens with 'stable portage tree' if it will born at some time! |
| 39 |
|
| 40 |
*** |
| 41 |
|
| 42 |
Maybe I'm completely wrong, but I think 'stable portage tree' topic isn't |
| 43 |
really about needs for MORE STABLE portage tree. I think it's about ppl |
| 44 |
who doesn't like regular updates, who update system only every 3+ months |
| 45 |
and who have troubles with such updates: too many packages changes at once, |
| 46 |
can't rollback to previous versions because they was deleted from portage, etc. |
| 47 |
|
| 48 |
For me - this is just because of portage nature: while it's possible to |
| 49 |
update when you want, doing this seldom make every update much harder and |
| 50 |
result in less stable system and feeling you've less control over system. |
| 51 |
|
| 52 |
I think this just should be documented as weakness (or just nature) of |
| 53 |
portage system: seldom (or no) updates result in less stable system! |
| 54 |
|
| 55 |
-- |
| 56 |
WBR, Alex. |
| 57 |
-- |
| 58 |
gentoo-server@g.o mailing list |
Archives