| 1 |
On 30/04/12 11:47, Michał Górny wrote: |
| 2 |
> On Mon, 30 Apr 2012 11:40:26 +0200 |
| 3 |
> Krzysztof Pawlik <nelchael@g.o> wrote: |
| 4 |
> |
| 5 |
>> On 30/04/12 11:30, Michał Górny wrote: |
| 6 |
>>> On Mon, 30 Apr 2012 10:24:58 +0200 |
| 7 |
>>> Krzysztof Pawlik <nelchael@g.o> wrote: |
| 8 |
>>> |
| 9 |
>>>> On 30/04/12 08:57, Michał Górny wrote: |
| 10 |
>>>>> Hello, |
| 11 |
>>>>> |
| 12 |
>>>>> Since lately Gentoo devs force you to replace collision-protect |
| 13 |
>>>>> with protect-owned [1] and sometimes packages just spit out files |
| 14 |
>>>>> randomly on the filesystem due to random errors, I thought it may |
| 15 |
>>>>> be a good idea to provide a new feature limiting the locations |
| 16 |
>>>>> where packages can install. |
| 17 |
>>>> |
| 18 |
>>>> You're not forced to do anything. |
| 19 |
>>> |
| 20 |
>>> I am not? How come random ebuilds die in pkg_pretend() for me then? |
| 21 |
>>> |
| 22 |
>>>> |
| 23 |
>>>>> In order to do that, we should first compose a complete |
| 24 |
>>>>> include/exclude list where packages can install. I'd suggest the |
| 25 |
>>>>> following: |
| 26 |
>>>>> |
| 27 |
>>>>> + /bin |
| 28 |
>>>>> + /boot (but maybe just subdirectories so packages can't overwrite |
| 29 |
>>>>> kernels?) |
| 30 |
>>>>> [potentially + /dev? but that's useful only when tmpfs isn't |
| 31 |
>>>>> mounted] |
| 32 |
>>>>> + /etc |
| 33 |
>>>>> + /lib, /lib32, /lib64 |
| 34 |
>>>>> + /opt |
| 35 |
>>>>> + /sbin |
| 36 |
>>>>> [potentially + /service for ugly daemontools] |
| 37 |
>>>>> + /usr |
| 38 |
>>>>> + /var |
| 39 |
>>>>> - /usr/local |
| 40 |
>>>>> - /usr/portage |
| 41 |
>>>>> |
| 42 |
>>>>> What are your thoughts on this? |
| 43 |
>>>>> |
| 44 |
>>>>> [1]:https://bugs.gentoo.org/show_bug.cgi?id=410691#c4 |
| 45 |
>>>> |
| 46 |
>>>> I think it's feature creep - you can just set |
| 47 |
>>>> INSTALL_MASK="/usr/local /usr/portage", no need for new features. |
| 48 |
>>>> |
| 49 |
>>>> PS. I (and few other folks) don't have /usr/portage. |
| 50 |
>>> |
| 51 |
>>> INSTALL_MASK won't make committing such an ebuild a fatal error. It |
| 52 |
>>> will just hide problems. |
| 53 |
>> |
| 54 |
>> How would your proposed feature prevent ebuilds that install |
| 55 |
>> something to /usr/local from being committed? |
| 56 |
> |
| 57 |
> If we enable this by default, devs will get errors for that. |
| 58 |
|
| 59 |
Could it be added instead to repoman? I'm testing all ebuilds before committing |
| 60 |
like this: |
| 61 |
|
| 62 |
$ ebuild xhtml2pdf-0.0.3.ebuild manifest clean install |
| 63 |
|
| 64 |
repoman could look at build.log or ${D} and warn if it can't do those checks. |
| 65 |
|
| 66 |
>> Other valid case are private repos - I have a few ebuilds that |
| 67 |
>> install stuff to /<some_random_dir> (proprietary stuff). |
| 68 |
> |
| 69 |
> Feel free to disable it. That's what FEATURES are for. |
| 70 |
|
| 71 |
Yes, I could. |
| 72 |
|
| 73 |
-- |
| 74 |
Krzysztof Pawlik <nelchael at gentoo.org> key id: 0xF6A80E46 |
| 75 |
desktop-misc, java, vim, kernel, python, apache... |
Archives