[QGIS-Developer] Thoughts on using processing models for ETL tasks

[Nyall Dawson]

On Wed, 27 Jun 2018 at 17:56, johnrobot <johnrobot at gmail.com> wrote:
>
> Continuing my quest for QGIS ETL perfection, I have a few more questions:
>
> 1) Is there a way of selecting output from multiple algorithms as input for
> an algorithm? In the drop down list, I can only select one input source and
> I often want to use more than one.

This works, if the algorithm accepts multiple layers. E.g. merge
vector layers, package layers, etc. It doesn't work for algorithms
which work on a single input layer (e.g. buffer).

> 2) I want to adjust lower/upper attribute string values. I can of course use
> upper/lower functions in the Field Calculator but what if I want to change
> "firstname lastname" to "Firstname Lastname"? Is there a way of changing
> only the first character in each word?
>
> 3) Is there a way of splitting an attribute with a specific character? For
> example, I might want to split an attribute with '>' and set an upper limit
> for how many new attributes may be created by this operation, in case the
> input attribute contains more than one '>'. This seems hard to do with the
> Field Calculator, where I have to specify the name and data type of a single
> new attribute.

Both these could be done by the "Refactor fields" algorithm.

> 4) Continuing on #3: Why do I have to specify the data type inside a model?
> Shouldn´t that be a task when writing the output? Before that, the user
> shouldn´t have to think about data types.

Can you elaborate?

Nyall