It looks like they are in order but maybe your data isn't where you think it should be vs. where it is? Zoom in to each cluster in this example and it does what you want. But each cluster is spaced far from each other.

Attached fmwt

Workbench I made real quick for example;

Column/Row for each set (similar to your screen shot)

Zoom in closer to see each set (2 of the 9 sets)

Zoom in on one to see the final result (one of the 9 sets)

You end up with the dissolved polygon for each of the sets of points. Let me know if this helps or not.

Attach your 4911_MS.txt file or export as a .fmwt

How to export a fmwt

http://docs.safe.com/fme/2020.1/html/FME_Desktop_Documentation/FME_Workbench/Workbench/saving_as_template.htm

***When you reply, reply to this message with the "Reply" button below and not the "write an answer" at the bottom. It will keep the messages in order.

Thanks for the instructions, sorry for the mess.

Here you are image attached

Don't think the .txt is in the .fmwt file. Make sure you hit the check mark to include it when you export the template.

Is it right now?

Can I ask you where are you from? I'm just curious...

Any way, I just don't understand why I must to "sorter" after the dissolver to have data in the right order (ascending). If I make workflow like--> dissolver+sorter(line_text_number or _row ascending)...It's ok! But why "dissolver" doesn't dissolve data in the right way immediately?

You can remove attributerounder before dissolver, It's useless....

Adding a sorter before the dissolver seems to fix the issue. Attached fmwt

I also tweaked a few things to give you some ideas of how to make it a bit easier to do what you are doing with far less transformers.

• Instead of using a creator to create your boxes, you can use a vertex creator with an X of 0 and a Y of text_line_number to get the same result

• You can do away with most of your transformers for parsing by splitting and using an attribute manager to retrieve the values. That will do away with most of your substring extractors.

New setup to replace all the extractors

AttributeManager

Result

Let me know if you have any questions.

Oklahoma City, Oklahoma.

The reason you had to sort again is because the data went through the feature merger. That transformer reorders the data when doing its comparison to merge. Changing over to the vertex creator and a buffer removes the need to sort due to the merger being removed.

Yes but it is anyway strange cause datas in the outport "merged" of the "feature merger" are in order ascending! I cannot see this change order in the feature merger.

I thank you very much for all these advices, to replace substring extractors and attributecreator/remover/keeper etc was the last passage make clean and performer my workflow, I was thinking about it. Thank you so much.

Giuseppe, Italian from Switzerland 😋