I'm trying to make a simple join in QGIS 3.10 between a table with a generated ID (using "Add unique value index field") named POINTID and a table created by the Distance Matrix algorithm with an InputID field (whose values come from the POINTID).

enter image description here

But this simply doesn't work as I get an error message during the processing saying that all my features could not be matched…

enter image description here

First, it looked like a duplicate of Joining attributes by field value not working but I double checked the fields type and it is the same: int/integer(0,0) as you can see in both the layers properties and in the logs.

I also checked that IDs are common to both tables so that the join can be done.

enter image description here

Strange thing but maybe irrelevant: the length of the generated field is 0. But the documentation of this function doesn't say anything about setting a field length (see enter link description here). The way the values are sorted in the table from Distance matrix is strange too (1,10,11…,2,20,21,…).

Also, it looks like the join works well on fields of type string, but I didn't find a way to generate unique ID fields of type string (I'm using the graphical modeler).

This gets me confused, any ideas what I'm doing wrong ?

[EDIT] : here is some simple csv data to check the process (SRID is 2154)

Table A


From table A, I generate a POINTID field using X column as class for algorithm "Add unique value index field".

Table B (one point from which to compute distance)


Table which is the result of Distance Matrix (distance values may be wrong but it doesn't matter):

  • To me, it is not really clear what you are doing and why. What I tried to understand, but not sure if I'm right: In fact, you do NOT want to make a table join, but to run a distance matrix algorithm from point layer_1 a to point layer_2, but the Unique identifier field for the two layers don't much. Right? However, your screenshot shows the algorithm Join attributes by field value. So I'm quite confused. Can you share a sample of your data?
    – Babel
    Commented Mar 26, 2021 at 22:15
  • @Babel Actually, I do want to make a join (with join attributes by field value) between the result of distance matrix algorithm and the input table. Let's say : from table A, I create a generated unique value index field. Then, I compute a distance matrix between table A and a table B (which contains only one point). Finally, I want to join the distance matrix with the original table A by keeping the distance result. Everything is done inside the graphical modeler. Is it possible to share data in SO ? Or should I open an issue on github ?
    – Beinje
    Commented Mar 27, 2021 at 16:05
  • Unfortunately, you can't share data here - if possible, load it to a cloud are somewhere.
    – Babel
    Commented Mar 27, 2021 at 16:16
  • Didn't find a good way to share shapefiles, but I edited my post and included some simple csv data (to load as new layer) if you want to reproduce the process.
    – Beinje
    Commented Mar 27, 2021 at 17:59
  • see my answer I posted before
    – Babel
    Commented Mar 27, 2021 at 18:00

2 Answers 2


I suppose you did not save the output of the distance matrix: it's a temporary layer, right? Replicating your workflow, I encountered the same problem as you as long as the layer was a temporary layer. If you save it as Geopackage, the problem disappears.

If you have just a temporary layer, the InputID of the output layer of Distance Matrix behaves strange: if you try to sort it by clicking on the column title, the sort order behaves as if it were a string: 1, 10, 11, [...], 2, 20, 21, [...], 3, 30, 31, [...], even though the field type is shown as integer. That's why the join does not work. I'm not sure if the problem is connected with the fact that Distance Matrix generates a multipoint layer, not a single point layer.

However, the solution is simple: as stated you can save the output (define an output path in the Distance Matrix dialog or make the created temporary layer permanent).

Another option (if you want it to keep only as temporary layer) would be to create a new field join_id using the field calculator with the following expression: to_int(InputID). With this new field, joining works perfectly.

  • You're right, I'm working with temporary layers, because everything is done through a model in qgis graphical modeler. Yet, my model needs to be used by final users so I prefer that they should not have to save intermediate layers. I guess I'm going to try the second option then. Btw, I think I will also submit this issue on qgis github because as you say, Distance Matrix seems to create a string field though it is shown as integer, so this is strange. Many thanks for your help !
    – Beinje
    Commented Mar 27, 2021 at 18:07
  • It would be great to report back if you find something out. In your case, I would use the approach in the other solution, using an expression. This is mire straightforward and avoids creating new layers.
    – Babel
    Commented Mar 27, 2021 at 18:48

There is even a direct way to create a new field in your layer, without the workaround of using Distance matrix. Let's say you have a layer points with many points and a layer center with just one point. In the layer points, create a new attribute with the distance of each point to the center using field calculator with this expression:

distance (
    geometry (get_feature_by_id ('center', 1))
  • Actually, my layer 'center' is itself a result of an algorithm in my model. Any idea how I should replace the explicit layer name (here 'center') by the ouput of the algorithm ? I tried @Algorithm_name_OUTPUT as it appears it the field calculator variables, but I get "Not root node, parsing failed ?!" during execution.
    – Beinje
    Commented Mar 29, 2021 at 9:07
  • FYI, it looks like your solution is not currently possible as of QGIS 3.16, see gis.stackexchange.com/questions/379549/… and github.com/qgis/QGIS/issues/30397#issuecomment-505699624.
    – Beinje
    Commented Apr 27, 2021 at 9:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.