[Qgis-developer] [Qgis-user] Any working GRASS in QGIS available? - Processing Bug?

[Alex Mandel]

Ah but we're closer to finding a solution with all those details. It may
be fixable in Processing, which can be pushed to the plugin repo for
update any time.

Passing this along to devs who might have enough information now.

Thanks,
Alex

On 08/02/2015 03:07 PM, Bernd Vogelgesang wrote:
> Am 02.08.2015, 22:44 Uhr, schrieb Alex Mandel <tech_dev at wildintellect.com>:
> 
>> After some brief testing, my OSGeoLive 8.5 VM with QGIS 2.6 and
>> Processing 2.9.3, GRASS 7 works.
>>
>> My QGIS 2.8, GRASS 6/7, Processing 2.10.1 does not.
> 
> Gave Windows another try: Removed everything, and installed the simple
> install OSGEO4W setup.
> QGIS 2.10.1 with Processing 2.10.1 and GRASS 6 working.
> 
> switched back to Ubuntu, cause of a bug in the modeler version of GDAL
> Dissolve Polygons (which is crucial for me).
> https://hub.qgis.org/issues/13174
> 
> Ubuntu: Now that I installed 2.8 with the debian repository only (so
> without ubuntis depencies), my GRASS 6 works (sorted out another error,
> that GRASS takes a column name "OR" from shape as an sql-command or
> sth), update to Processing 2.10.1 worked as well.
> 
> The drawback, no SAGA, which might come handy cause the now available
> algos do not really do what I expect ;)
> 
> I think the problem is within Processing in combination with the packaging:
> When e.g. I install QGIS 2.8 with ubuntugis-ltr dependencies (where
> nobody on the install page claims I shouldn't do), GRASS7 and the
> modeler are working, but Processing version is 2.6. Updating Processing
> to current Processing 2.10.1 (as recommended) seems to work.
> (Besides that v.clean in the modeler only returns a polygon layer from
> an polygon input, if I also set an output name for the error layer,
> otherwise I get an empty line shape)
> 
> But: The next time I run QGIS and want to run an GRASS algo: Missing
> depencies
> 
> I can now replace Processing 2.10.1 with the second latest 2.9.3 from
> https://plugins.qgis.org/plugins/processing/version/2.9.3/
> and there v.clean does what it should in Processing AND modeler.
> 
> But: Now I do not have GDAL Dissolve polygons which is in Processing
> 2.10.1. Well, I could switch back to Windows QGIS 2.10, but, grr, there
> is the bug in this algo ...
> 
> You see, I'm trapped.
> 
> Thanx for you attention
> Bernd
> 
> 
> 
>>
>> Both are Ubuntu 14.04 based.
>>
>> I'm not sure if this is a bug in Processing, or in packaging. I recall
>> in OSGeoLive 8.5 that we had custom packages with a patch. Note, you can
>> get those versions for your Ubuntu system from
>> https://launchpad.net/~osgeolive/+archive/ubuntu/release-8.5
>>
>> We freeze a copy of the working versions in our own ppa for long term
>> history.
>>
>> Thanks,
>> Alex
>>
>>
>> On 2015-08-02 03:32, Bernd Vogelgesang wrote:
>>> Hi Alex,
>>>
>>> Am 02.08.2015, 01:22 Uhr, schrieb Alex Mandel
>>> <tech_dev at wildintellect.com>:
>>>
>>>> It's not a Windows/Ubuntu/Mac/OS thing, it's a QGIS + GRASS version
>>>> change thing.
>>>>
>>>> Please indicate which version of QGIS and GRASS you are trying to get
>>>> working.
>>>
>>> Actually I do not mind any more what version I would like to get
>>> working. I would have preferred the 2.8-LTR branch with no-matter-what
>>> GRASS version.
>>>
>>> I just would like to be able to finish my project, and there is
>>> unfortunately no alternative to the GRASS algos in my model.
>>>
>>>> QGIS 2.6 and older works with GRASS 6, 2.8+ sometimes is built to work
>>>> with GRASS 7 (though possibly also with GRASS 6).
>>>
>>> Mhm, reverting to 2.6 ... Lets see if this still possible at all ...
>>>
>>>> If you want a guaranteed working setup ASAP download a copy of OSGeo
>>>> Live Virtual Machine, we made sure that it works in that.
>>>> http://live.osgeo.org
>>>
>>> Will give it a try. Download will take another 1.5 hours ...
>>>
>>> So does that mean that even your do not have a working GRASS in a recent
>>> QGIS by "normal" means but the osgeo VM?
>>>
>>>
>>>>
>>>> Thanks,
>>>> Alex
>>>
>>> Cheers
>>> Bernd
>>>
>>>>
>>>> On 2015-08-01 14:41, Bernd Vogelgesang wrote:
>>>>> Hi guys,
>>>>> is it just bad karma, that I seem not to able to get a working
>>>>> GRASS in
>>>>> QGIS? Or did I miss the big advertisement "GRASS out of order"
>>>>> somewhere.
>>>>>
>>>>> For nearly 2 weeks, I tried to get things to work in Ubuntu/Mint, and
>>>>> tried every possible/impossible repository, but always GRASS has
>>>>> missing
>>>>> dependencies, the forms open in Processing, but the algos don't run
>>>>> through, or the functions are working in Processing, but failing in
>>>>> the
>>>>> modeller.
>>>>>
>>>>> As I really got to finish a model in the Modeller including some GRASS
>>>>> algos, I finally turned on my VM with Windows ...
>>>>> but what the heck? Same s...t there!
>>>>>
>>>>> I really wonder why there aren't many post about trouble with GRASS on
>>>>> this list or at stackexchange, normally this would explode when things
>>>>> like these happen.
>>>>>
>>>>> So am I really the one with the bad karma, do you all other simply not
>>>>> use GRASS in processing as much as I do, or do you all simply have
>>>>> working installs, so there is not problem?
>>>>>
>>>>> Please, at least one: Do you have an install under Ubuntu or Windows,
>>>>> where GRASS algos are functional in the Modeler ? If yes, which
>>>>> version/what did you do?
>>>>>
>>>>> ... Soon, I will go and find me a rope ...
>>>>>
>>>>> Bernd
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
> 
>