[Qgis-developer] backports for 1.7.3

[Sandro Santilli]

On Wed, Nov 30, 2011 at 09:38:12AM +0100, Giuseppe Sucameli wrote:
> 
> On Nov 30, 2011, at 8:50 AM, Sandro Santilli <strk at keybit.net> wrote:
> 
> > On Wed, Nov 30, 2011 at 12:01:40AM +0100, Giuseppe Sucameli wrote:
> >> On Tue, Nov 29, 2011 at 11:08 PM, Sandro Santilli <strk at keybit.net> wrote:
> >>> What event could make it impossible to release 1.7.whatever ?
> >> 
> >> Is QGis reputation enough? ;)
> > 
> > I don't see how releasing early and often may degrade QGis reputation.
> 
> I don't meant reputation is related to realeasing early itself, but if early means bugged...

We're talking about a stable branch here, so _theoretically_ in NO event
should a "dangerous" patch be applied there. Only bug fixes should.
This is the whole point of stable branches.
You force yourself to carefully consider any modification as for its safety.
Every single patch could trigger a new bugfix release (if it was so cheap).

> I agree with releasing early for releases from the master branch only because that one is continuously tested.

How so ? By developers, you mean ?
Those testers don't even need a release, they just get the code from repository,
and anyone can get a daily snapshot from github or wherever it is.

It'll be _very_ important to have real "continuos testing" performed by
machines over all branches, and particularly over the _stable_ branches.

> Users don't like to upgrade often and sometimes they want it but cannot.

Let them wait until 1.7.10 ... it'll be their choice.
But for the rest of us, why waiting ?

I don't like to upgrade either, when it is a major upgrade (things can break).
But bugfix upgrades could be painless, if done carefully.

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html