Should hidden files and directories be treated differently by the home plug

[Neil McPhail]

On 06/07/16 22:52, Neil McPhail wrote:
> On 06/07/16 22:46, Mark Shuttleworth wrote:
>> On 06/07/16 17:41, Neil McPhail wrote:
>>> On 06/07/16 18:13, Mark Shuttleworth wrote:
>>>> In principle I think the way to do this is to have the snap declare 
>>>> some
>>>> specific .files that it wants to see, as parameters on the interface.
>>>> This would trigger a one-time review, but assuming the request is 
>>>> sane,
>>>> those would then be granted.
>>>>
>>>> The silversearcher example, though, is more interesting because it's
>>>> explicitly wanting to search dotfiles that belong to other apps.
>>>>
>>>> Mark
>>>>
>>> If a solution could be found for silversearcher, it may become a class
>>> solution for useful snaps of "git" and other build tools. I'm sure we
>>> need a more generic solution for dotfiles than exclusion+whitelisting.
>> Why would a git snap need to see dotfiles other than its own?
>>
>> Code trees are not in .foo, they are in places like src/foo/ and
>> dotfiles inside those would of course be readable using the home 
>> interface.
>>
>> Mark
> Not everyone organises their git repos without dotfiles ;)
>
> For example, I'm currently hacking on the neverball-touch repo which 
> has a ".build" file and ".tx" directory in the root. A snapped version 
> of git, for example, wouldn't see them.
>
> NMP
>
Sorry, ignore the noise. You are correct, of course. I hadn't 
appreciated dotfiles in subdirectories are treated differently by the 
home interface.

Apologies

NMP