[Qgis-developer] Data-defined symbol size is not backwards compatible to 2.8.2

[Andreas Neumann]

I found another issue with the new scaling method:

For multi level point symbols we need to be able to define the scaling 
on the symbol level. However, when adding it on the symbol level, the 
same rule applies both on the parent marker and on the individual symbol 
level.

What if one part of the symbol should be scaled differently than the 
other? We really need this to work on the indivual symbol levels, not on 
the parent marker - or at least make it optional that it scales the 
parent as well.

As it works currently, I get a very different rendering in the map, 
compared to the preview in the legend ;-(

Andreas

On 02.06.2015 15:53, Andreas Neumann wrote:
> Hi,
>
> Thanks for explaining. The assistant is very nice and a welcome 
> improvement! Thanks for investing and working on it.
>
> I had an issue that I had a size defined in 2.8.x with "scaling by 
> area" which, when opened in QGIS master displayed much bigger. Once I 
> changed that in 2.8x to "scale by diameter" the symbol sizes are 
> identical in master.
>
> So there seems to be an issue with converting "scale by area" symbol 
> sizes from 2.8 over to QGIS master.
>
> Should I open a bug report or is one open already?
>
> Thanks,
> Andreas
>
> On 01.06.2015 20:05, Régis Haubourg wrote:
>> Hi,
>> here are the ideas behind this work, Nyall (code reviewer) and Vincent
>> (author) could explain implentation choices more than me (funder):
>>
>> - get a more consistent UI with data defined widgets, and not "advanced
>> fields". That way, size is in one place only.
>> - offer an assistant on size varying common expression. You will find 
>> it at
>> the bottom of the drop down widget. It computes max value from field or
>> expression and allows normal user to do what other GIS do.
>> - That assistant offer legend previsualisation, and generates a 
>> legend for
>> map and composer. I wish we have a legend for any expression...
>> - During implementation, we understood that symbol size was a 
>> multiplication
>> factor of size varying factor. That implied that it was impossible to
>> predict final screen size. The new implementation clears that up.
>> - offer a size varying graduated renderer, allowing the use of
>> classifications algorithms
>> - offer a legend for diagrams (yes!)
>>
>> IMHO, we need to read previous versions correctly, but that sometimes 
>> need
>> to read all features to retroengineer a size expression. Vincent have
>> planned to polish that now that feature freeze is made.
>>
>> All that needs testing of course. I'm not totally satisfied with the
>> assistant shortcut, hidden inside the size varying widget. If someone 
>> have a
>> better UI idea..
>>
>> Hope that helps clarifying those changes.
>> Cheers
>> Régis
>>
>>
>>
>>
>>
>> -- 
>> View this message in context: 
>> http://osgeo-org.1560.x6.nabble.com/Data-defined-symbol-size-is-not-backwards-compatible-to-2-8-2-tp5208256p5208525.html
>> Sent from the Quantum GIS - Developer mailing list archive at 
>> Nabble.com.
>> _______________________________________________
>> Qgis-developer mailing list
>> Qgis-developer at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/qgis-developer
>
> _______________________________________________
> Qgis-developer mailing list
> Qgis-developer at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/qgis-developer