[Qgis-developer] [Fwd: Re: [Qgis-user] Poll on website]

[Alex Mandel]

I agree with Tim and will point out that in my experience 1.0.x is still
more stable. Most of my colleagues that I see on a daily basis still run
it, and I still have a copy installed on 1 machine at all times because
the upgrade path has not appeared clear and tested, nor did it work when
1.1 first came out. Previous upgrade attempts on both Ubuntu and Windows
are really untested and don't go that well. (GRASS-GDAL plugin always
seems to break)

For this upcoming school year I plan to request 1.0.x be installed on
our lab machines (40+) since I know the functions we need are working. I
can likely request minor updates but the lab managers aren't keen to big
jumps in versions every 3 months, since there's lots of other software
they have to play nice with. So +1 on getting the exe installers working
better out of the box (I'm experimenting with msys builds to see if that
helps)

I think one thing we need to start focusing on more is testing, like
before 2.0 we need to test that an upgrade to 2.0 can be done from 1.0
with OSGeo4w and on Ubuntu, the dependencies on GDAL and GRASS seem to
make these unstable.

I'm also interested in getting more discussion going about automated
testing to find bugs before release.

Thanks,
Alex

Mailing Lists wrote:
> Hi All
> 
> I'm not in agreement with ditching 1.0.x. The stats we looked at
> (including OSGEO downloads) for the win setup.exe's that I made for
> 1.0.2 and 1.1 and made available at exactly  the same time had many
> many more downloads for the stable release (~75k for ver 1.0.2 versus
> ~4.7k downloads for 1.1). True we cant see how many downloads were
> made on other distros, third party redistribution etc,  but if the
> figures lay the other way around wouldnt you be using them as an
> argument for deprecating 1.0.x?
> 
> I would like to make the additional point that many users using the
> 1.0.x releases are probably silent on this list, the forums etc - this
> discussion is biased towards developers who typically enjoy being on
> the bleeding edge.
> 
> Surely it does not impact on any of you if 1.0.x continues to exist
> and the users who are looking for a long term maintained version will
> appreciate our commitment to not only making new versions, but also
> providing a slow moving target for them to work against.
> 
> That said I suspect this thread will go on ad-infinitum since so would
> like to suggest that we do the following:
> 
> - Promote the 1.x releases on the download pages etc
> - Keep the 1.0.x description and down load page there (in a less
> prominant position).
> - Do occasional 1.0.x releases  - I will backport any commits I can to
> the 1.0.x branch and make occasional releases
> - Release 1.2 with the current feature set as Marco and others have suggested
> - Incorporate Martins NG and Labelling stuff into the future 1.3 release
> - Lets work to get some kind of proper metrics in place - an opt in
> first-run wizard that is invoked each time a user uses QGIS + unique
> version that pings a server with OS + Version. I will do the server
> side stuff if someone else can pitch in with the QGIS part.
> 
> 
> 
> Regards
> 
> Tim
> 
> On Sat, Jul 18, 2009 at 1:20 PM, Werner Macho<werner.macho at gmail.com> wrote:
> Hi!
> 
> I also fully agree - we should bring out 1.2 soon and probably drop 1.0.
> But the (most asked) thing is that we should have a fully working
> stand alone installer for windows.
> 
> Not that I like Windows (I only _have_ to use it @work) but it is
> still a fact that most people unfortunately are just able to click
> "Setup.exe"
> Dont' get me wrong - I like those people .. They are our Users ;) (and
> they give me work by cleaning their Computer)
> 
> So I still vote for getting 1.2 out soon but .. to get really more
> users to use it - we need Setup.exe ;)
> 
> regards
> Werner
> 
> Paolo Cavallini wrote:
>>>> Andreas Neumann ha scritto:
>>>>
>>>>> One of the big advantages QGIS currently has, is that I (as a
>>>>> user/customer) can get in new features relatively quickly (compared to
>>>>> commercial GIS). QGIS should use this advantage and release often.
>>>> Agreed fully. I think we cannot trust download stats at all, as many
>>>> (most?) people download qgis from OSGeo or their distro sites.
>>>> Furthermore, the prominent notice on our web site "1.0.2 is the only
>>>> good one" pushes hard downloaders towards that version.
>>>> Incidentally, I must say that having people using 1.0.2 makes also
>>>> people reporting old bugs as new ones, which makes keeping the bug queue
>>>> clean more difficult.
>>>> I repeat mysel, but I think that deprecating 1.0.2 and releasing 1.2 now
>>>> is the way to go. Please note that we only have 9 critical bugs open
>>>> now, the lowest number since ages.
>>>> All the best.
>>