| 1 |
On Thu, 2019-04-25 at 16:07 -0400, Michael Orlitzky wrote: |
| 2 |
> On 4/24/19 8:53 AM, Michał Górny wrote: |
| 3 |
> > systemd.eclass has more than one purpose, and therefore such dep didn't |
| 4 |
> > belong there (ebuilds should take care of the dependencies when using |
| 5 |
> > tmpfiles logic from it). tmpfiles.eclass on the other hand has a single |
| 6 |
> > purpose, so we've solved the problem by adding an RDEPEND there. |
| 7 |
> > |
| 8 |
> |
| 9 |
> I don't quite buy that either: do we assume that users will be running |
| 10 |
> either OpenRC or systemd? |
| 11 |
> |
| 12 |
> If we do, then the RDEPEND on virtual/tmpfiles is redundant. |
| 13 |
> |
| 14 |
> If we don't, then someone running (say) daemontools might want to |
| 15 |
> install a package that comes with a tmpfiles.d entry. Per our small file |
| 16 |
> policy, the tmpfiles.d entry should be installed unconditionally. And if |
| 17 |
> the tmpfiles entry needs to be processed immediately on OpenRC/systemd, |
| 18 |
> then tmpfiles_process() from tmpfiles.eclass must be used to support |
| 19 |
> OpenRC users. But, that will pull in virtual/tmpfiles unnecessarily on a |
| 20 |
> system that's running neither OpenRC nor systemd. |
| 21 |
> |
| 22 |
|
| 23 |
Wrong. tmpfiles_process() requires virtual/tmpfiles on any system, |
| 24 |
including daemontools, bare minimal init and whatever. Keeping it |
| 25 |
installed afterwards is a minor side effect, and technical limitation of |
| 26 |
our dependency types (lack of install-depend). |
| 27 |
|
| 28 |
-- |
| 29 |
Best regards, |
| 30 |
Michał Górny |
Archives