[QGIS-Developer] Call for possible vote on change to official plugin repo policy?

[Nyall Dawson]

Hi all,

I'd like to raise the possibility of putting to vote changes to the
official plugin repository.

This is something which I've been mulling over for a long time now,
and having discussed with other members of the community I believe
that there is sufficient interest to at least put this out as a
possible vote. It's a difficult topic, yet I hope that we can keep
discussion civil, professional and impersonal. Certainly no offence is
intended here -- I'm just after the best outcome for our users.

Here's the proposal:

"Should the official QGIS plugin repository flag plugins which have
known serious issues, and hide these plugins from users unless a 'Show
plugins with known issues' checkbox is ticked?"

Here's the rationale behind this:

There's a number of known badly behaved QGIS plugins, which despite
long-standing, very serious bugs, the plugins still are widely
installed and recommended by lots of books/blogs/users/training
courses/etc (who are likely unaware of these issues). Stackexchange is
littered with users encountering issues with these plugins, and I'm
constantly encountering users with issues in their QGIS installs
during my consultancy and training which is caused by these plugins.

I believe we have a responsibility to our users here -- if we are
aware that a certain plugin causes serious issues, why are we making
it available to users to download easily through our official plugin
repository?

The fact is: if we had bugs in core QGIS code which caused serious
issues like loss of user data, made QGIS randomly crash or slow to a
crawl, caused inaccurate measurements, or incorrect map rendering,  we
would consider those extremely serious blocking bugs and do everything
in our power to fix them. Yet we have plugins which we know DO cause
issues like this, and we make them accessible to users to install with
a couple of clicks and no warning.

It's a tricky and delicate issue... I fully understand that
significant development time and costs have gone into these plugins,
and that there is significant visibility benefits for the companies
behind having those plugins in the official repo, but the truth is
that the seriousness of these issues has unfortunately caused tangible
harm to QGIS' reputation for some users.

This topic is bound to be compared to Apple's "walled garden" approach
to their app store, which (in MY opinion rightly) comes under much
criticism for restricting user freedom. However, the situation is
quite different here because we supply all the tools required for
developers to make their plugins installable in many different ways
outside the official repo. Developers can create custom repos, they
can reuse the plugin library code on their own server, and users can
easily install non-repo plugins via the new "install plugin from zip
file" option in 3.0. Hiding or removing a plugin from the official
repo does NOT equate to making this plugin unavailable for QGIS users.

So my proposal is that:
- if a plugin has known, serious issues we contact the plugin author
to notify them
- we add a flag to the plugin to indicate it has known issues
- these plugins are not shown by default - users must opt in via a
checkbox (much like the "show experimental" check box)
- if/when the plugin fixes the issues, the flag is removed

Like I said at the start of this email... I realise it's a delicate
matter. But I strongly believe that our responsibility here lies with
protecting our users, not solely our developers, and we should take
some action to do this. That's why I'd like this raised to a vote with
the whole voting community vs just having discussion amongst the
PSC/developers themselves.

Nyall