[QGIS-Developer] Some thought on LTR

[Matthias Kuhn]

On 8/5/19 1:00 AM, Nyall Dawson wrote:
> On Fri, 2 Aug 2019 at 18:13, Matthias Kuhn <matthias at opengis.ch> wrote:
>
>>     d) Stop shipping bugfixes (hint: joke, that makes the LTR concept
>> pointless and anyone can do that today already by sticking to the .0
>> patch release ;) )
> Actually - jokes aside - this does raise a good question.
>
> We've never (as far as i know) formally defined about what the goal of
> the LTR is. Is it:
> 1. a version of QGIS with every bug fix possible backported
> or
> 2. a version of QGIS with only absolutely critical bugfixes
> backported, such as security risks or data corruption bugs

That's a very good question.

While there's no binary crystal clear categorization possible anyway, I 
realize that we tend toward 1. if it's a fixes affecting 
(experienced/sponsored/implemented by) ourselves and 2. for any other 
fix. Which introduces a bias that can cause tension.

So I agree that:

IF the concerns with the current (1.) approach are too high, the 
decision for 2. is better taken as a community and made official so it 
can be communicated easier to our clients and users.

(And then even with the decision taken, the situation will be tricky. At 
the time a client reports a bug and funding is available, it almost 
always looks like a trivial fix and it's easy to promise a backport 
"what could potentially go wrong?". Whereas the question about the 
categorization of a bugfix mostly appears in retrospective when it turns 
out something has gone wrong and the situation is already difficult.)

Matthias

> (or somewhere between the two)
>
> Currently, it's very much 1. We see everything backported from crash
> fixes to string updates to performance optimisations to backported
> API. Maybe this is the problem. Maybe we should only be accepting
> absolutely mission critical bug fixes, and the expectation is to see
> only 1 or 2 commits between LTR patch release versions. Reality is
> that every bug fix, regardless of how trivial it seems, brings with it
> the increased chances of regressions into the stable LTR release...
>
> Possibly this is a question we need to raise with our voting panel or
> user communities, in order to work out exactly what people's desires
> from the LTR are.
>
> Nyall