On Sat, Oct 18, 2014 at 10:51 AM, Rahul Sundaram metherid@gmail.com wrote:

Hi

On Sat, Oct 18, 2014 at 10:22 AM, Michael Catanzaro wrote:

...

On Sat, 2014-10-18 at 06:17 -0400, Rahul Sundaram wrote:

...

Would it be feasible to require that any default applications in workstation follow the appropriate freedesktop specifications including xdg base dir in

https://fedoraproject.org/wiki/Workstation/Guidelines/Applications_and_Launc...

This isn't the right policy for that, since that policy only covers apps with visible desktop files.

Fair enough. I am not particular about the location of the policy. Any comments on the suggested policy itself?

We didn't get to this in the meeting today, so we'll try and work this out over the list.

Personally, I have no major concerns with requiring default apps to support xdg configuration dirs, particularly since most already do. However, I don't think it's a pressing concern for F21 and most of the work should happen in the upstream projects. That is particularly true for Firefox and NSS/NSPR. The rest seem like quick enough work, but we might wish to grant exceptions for those 3 if we do adopt such a policy.

What do other WG members think?

josh

On Wed, 2014-10-22 at 16:06 -0400, Rahul Sundaram wrote:

    What do other WG members think?

While I focused on xdg base dir spec, I would like WG members to look at which freedesktop.org specs are at a mature stage and require them along with xdg base dir. freedesktop.org itself doesn't make that very clear.

I personally think that this entire angle of 'approval/exception' for 'allowing' applications onto the Workstation is a bit misguided, in particular for something like this.

It is worth keeping the overall goal in mind: We want to offer high-quality applications to the Workstation users - as many as possible.

It is one thing to write guidelines about best practices for logos, icons and such - those are things that can be fixed up in packaging.

Changing the location of config files on the filesystem is much more involved, and really needs upstream acceptance of the change. Otherwise, you're invalidating all the documentation that talks about configuration, and information that our users find on the internet will confusingly not apply to the Fedora fork of the app.

Therefore, I think we can't be too strict about this; we can certainly recommend this as best practice, and continue pushing for this as we've already done here: https://wiki.gnome.org/Initiatives/GnomeGoals/XDGConfigFolders But I don't think we can exclude apps based on this, if we want to be serious about having the best-of-breed applications available.

----- Original Message -----

From: "Rahul Sundaram" metherid@gmail.com To: "Discussions about development for the Fedora desktop" desktop@lists.fedoraproject.org Sent: Thursday, October 23, 2014 4:19:27 PM Subject: Re: freedesktop specs

Hi

On Thu, Oct 23, 2014 at 12:10 PM, Matthias Clasen wrote:

It is worth keeping the overall goal in mind: We want to offer high-quality applications to the Workstation users - as many as possible.

It is one thing to write guidelines about best practices for logos, icons and such - those are things that can be fixed up in packaging.

Changing the location of config files on the filesystem is much more involved, and really needs upstream acceptance of the change

IMO, all of the changes we make have to be send upstream anyway regardless of how minor it is. There is no particular reason a better logo shouldn't be upstream to use your example.

You say that, but Jakub Steiner tried to submit a much improved icon to the Darktable project not long ago and it got rejected with the suggestion that if we didn't want the current icon we should just patch it ourselves in Fedora :)

There should be a high level of consistency

for the default apps and that goes beyond just minor differences in packaging. Would you want to introduce a new default app that uses GTK2 at this point? If we say all default apps in Fedora workstation MUST use a standard toolkit like GTK3 or Qt4, that doesn't mean we will be forking the apps. Just that we wouldn't accept it into the default set until it is ported over. XDG base dir spec could be enforced in the same way.

I would be fine with that, but mostly because the likelyhood of anyone even considering adding such an app to the default installation is extremely low. The goal here is to not include as little as possible with the default set and instead rely on Software to be the portal to applications.

Christian

Hi

On Fri, Oct 24, 2014 at 11:34 AM, Christian Schaller wrote:

...

IMO, all of the changes we make have to be send upstream anyway

regardless of

...

how minor it is. There is no particular reason a better logo shouldn't be upstream to use your example.

You say that, but Jakub Steiner tried to submit a much improved icon to the Darktable project not long ago and it got rejected with the suggestion that if we didn't want the current icon we should just patch it ourselves in Fedora :)

Sure. I didn't want to get into all the possible scenarios here but we should always offer patches to upstream, no matter how minor they are. I try to go as far as reporting permission issues, empty files etc just because I think that is the right approach as outlined in https://fedoraproject.org/wiki/Staying_close_to_upstream_projects and endorsed by FESCo etc. That doesn't mean all our patches will be accepted. For practically all major changes, some patches will end up being carried downstream even for security issues.

My point was merely that considering that as our default mode of working, we are not advocating for forking just because we say that we will enforce a specific number of specs from freedesktop.org whatever they might be.

I would be fine with that, but mostly because the likelyhood of anyone even

considering adding such an app to the default installation is extremely low. The goal here is to not include as little as possible with the default set and instead rely on Software to be the portal to applications.

Agreed with the general idea of keeping the number of pre-installed components to a smaller set but there are some very high profile applications/components already included that don't follow the xdg base dir spec for example (firefox, nss, java etc) and by specifying that we are going to atleast except that any software components included by default follow the specs, we get to a more clearer definition of what we consider "high quality". If I am a software developer developing something for Fedora workstation, I would also gain a clearer understanding of the expectations involved.

In the case of xdg base dir spec, it serves a functional purpose of making backups easier and it isn't merely cosmetic. From a somewhat selfish perspective, I would also likely get less resistance upstream if a major distribution like Fedora says "This is the minimum we expect for any default components".

Rahul