Request: Ability to re-order objects in model _WITHOUT_ swapping object IDs

[jhardin-accumula]

Feature description
Changing the order of generation of objects by swapping object IDs has massive unwelcome side effects, as it changes the order of creation of both objects. This easily corrupts the model and causes problems when trying to reload it.

In addition to swapping IDs, it would be very welcome to have a way to relocate an object to a different place in the order of objects without changing the order of any other objects.

Additional info
Use case: create a collation and apply it to existing columns in the model, in v1.2.1

The collation is created at the very end of the model, as the last object (which is reasonable for a new object).

If you then apply that collation to a table column, you get a sequence error in Validation because the table definition appears in the model before the collation definition; if you Fix it the objects apparently get swapped rather than just the collation object being relocated to before the table object. This fixes the collation sequence error, but breaks relations to the table as the table is now at a different location in the model.

This may not be apparent immediately - it will validate successfully and save, but if you then reload the model you get errors regarding references.

This caused me to lose quite a bit of work because I had applied the collation to a dozen or more columns scattered around the model before saving and trying to reload the model.

Rather than trusting the repair process I decided to revert my model to a saved version and redo things. After creating the collation I saved the model and then edited the model file to move the collation definition to near the top, without touching any other objects.

It would be wonderful if there was a way to do this via the UI.

It would also be great if the Validation Fix for sequence errors moved only the referenced object to earlier in the model, rather than apparently swapping the referenced object and the referencing object.

[rkhaotix]

Hi @jhardin-accumula

That's strange because the validation, even swapping objects' positions, fixes all reference problems if you let it run until no more issues are reported. Maybe you have stumbled on a bug. Is it possible for you to send me a copy of your database model? It can be the broken one (if you still have it) or the functional one. I want to understand why the validation isn't doing what it's designed to do in your case.

I suspect that it's a problem specific to collations, but I'll investigate carefully. I'm finishing the first alpha of pgModeler 2.0 (to be released by the end of this month), and after that, I will take care of bug reports. :)

Thanks.

[jhardin-accumula]

Raphael: I went through trying to set up a repro for you and the behavior changed... I defined the collation and added it to one column. Validation failed, and I fixed it and saved, and then reloaded, and the model loaded without errors. When I was doing this the first time I added the collation to existing columns on several different tables, with validations in between at somewhat random intervals as I worked through the model. It's possible that the corruption occurred after *several* "add collation to column(s) -> validate -> fix errors" cycles swapped multiple things around. I did find it quite odd that every validation -> fix cycle did pass validation, but that saving the validated model couldn't be re-loaded without error... I will try to repro this, but I can't promise any time window for delivery or guarantee that I will hit the needed sequence of operations to repro the problem. :( I will also have to get approval to provide a copy of the model.

[rkhaotix]

No worries!

I'll investigate it anyway. Also, maybe I change the way pgModeler change the objects places.

I'll focus on big fixes and new features once I finish my work on the refreshed UI for pgModeler :)