Blueprint Assist Plugin

[Antidamage]

I'd like to propose a concept: fences.

They'd be an invisible rectangular boundary around event node trees in a graph. Blueprint Assist would be able to ensure that every node tree was inside its own non-overlapping fence. I kind of picture it as being a bit like a grid widget, in that it would automatically form rows or columns.

The new behaviour is that nothing would overlap by starting at the top-left most node and working down and across from there. Any nodes not anchored to an event graph should be moved into their own fenced area down the bottom.

The use case is that node trees quite often end up overlapping when formatted and it's quite tedious to continually nudge everything out further and further. Even when they don't overlap, it would be nice to have a good bit of space between separate node trees. It would be very nice if the plugin just did that for you. I also quite often get spare node networks that are not connected to any event sitting behind or in front of my active node trees. It's like picking apart a knot separating them out.

I know some similar functionality already exists in the plugin but I wasn't sure if anything does exactly this.

[fpwong]

I'd like to propose a concept: fences.

They'd be an invisible rectangular boundary around event node trees in a graph. Blueprint Assist would be able to ensure that every node tree was inside its own non-overlapping fence. I kind of picture it as being a bit like a grid widget, in that it would automatically form rows or columns.

The new behaviour is that nothing would overlap by starting at the top-left most node and working down and across from there. Any nodes not anchored to an event graph should be moved into their own fenced area down the bottom.

The use case is that node trees quite often end up overlapping when formatted and it's quite tedious to continually nudge everything out further and further. Even when they don't overlap, it would be nice to have a good bit of space between separate node trees. It would be very nice if the plugin just did that for you. I also quite often get spare node networks that are not connected to any event sitting behind or in front of my active node trees. It's like picking apart a knot separating them out.

I know some similar functionality already exists in the plugin but I wasn't sure if anything does exactly this.

If I understand correctly, these fences would apply when formatting nodes but not when dragging nodes around the graph (I also don't know how to get a callback from dragging nodes currently). Basically if a node tree were to be overlapping after formatting it would then move into a good location on the graph so it avoids collision.

This behavior (with a grid) is pretty much how the "format all" command works currently. It lays out all the node-trees from top to bottom using the bounds of each node-tree. As you said loose node-trees (with no event node) can still overlap though. I'll add an option to move these out of the way in the next major update. Give this command a try if you haven't yet, personally I use the format-all command a lot instead of formatting one node-tree at a time.

As for the fences, I'm not sold on the feature snapping to a grid, since you might as well use the format all command. But it would be nice to not overlap other node-trees after formatting. I'll look into adding this behavior in a future update.

[Antidamage]

That delay is alright, I've been preferring to manually hit the format button when I'm ready rather than have stuff disappear on me.

Would it be possible to add an option to format-all to specify how much space to put between node trees?

I didn't mean it would snap to a grid, I mean it would act like a grid component in UMG and align fences either in columns or rows with the user choosing which behaviour is dominant. But you get the general idea. I can do a diagram if you like.

[fpwong]

Antidamage Have a look at the "Format All Padding" setting.

The issue I have with the UMG grid behavior (I might be wrong here) is that functions in BP have a large variation in length, so your rows get really long which causes a lot of empty space. Which is why the format all command runs in a single column.

I could try to adjust the formatting algorithm so that it is able to wrap a function once it gets past a certain width. This would likely solve this issue so then you could actually fit the function into any width row you want to define. (This wrapping feature does sound very useful too, will be adding this eventually too).

I've been preferring to manually hit the format button when I'm ready rather than have stuff disappear on me

Also could you clarify what you mean by this? The format all command should work the same by pressing 'CTRL + R'.

[Antidamage]

Yeah, that's kind of the point. It's nice to have where they start all lined up. Then the same for the next column over, no matter how big the previous column was. I don't mind zooming out and dragging across, in fact that's kind of the point.

How about this behaviour: the plugin detects what event nodes are almost in a column (say within a thousand pixels) and aligns them. Then we can continue to separate logic groups by column.

> Also could you clarify what you mean by this? The format all command should work the same by pressing 'CTRL + R'.

That's what I do. I mean that I don't use auto-format because half the time the node I created disappears from where I created it.

[TobyTheKiwi]

fpwong Hi

I'm seeing a bug at the moment where use CTRL + SHIFT + R (refresh cache) and it causes a comment to add a whole set of nearby nodes that weren't included in it before the refresh. I've included some images to try illustrate the behaviour.

I tried manually resetting the comment to the correct nodes but it will repeatedly include the neighbouring comment and nodes.

Obviously not the end of the world but still rather frustrating behaviour.

Thanks!

[fpwong]

TobyTheKiwi I have noticed a similar issue recently when undoing formatting, will try to fix this asap.

Antidamage I'll see if I can add an option to the format all command to support this behavior.

[Antidamage]

This is why this is my favourite asset on the marketplace. Cheers man

[fpwong]

TobyTheKiwi The bug with comment nodes seems to be related to the "Use Knot Node Pool" setting (at least for undo). Still working on a fix but try disabling this setting for now.

[fpwong]

Update 9 May 2020 (v2.3.1):

• Added support for 4.25
• Fixed crash when opening AddSymbolMenu with no valid widget

[TobyTheKiwi]

fpwong
I think this might be introduced behaviour recently but I'm not sure. I'm seeing this sort of reroute node where it places it much further to the left and crossing the wire under nodes.

[TobyTheKiwi]

It also seems like the logic that makes switch statements fan out nicely might be arranging false branches with only a single node much further away than they need to be.

[fpwong]

TobyTheKiwi Ah yes, that is indeed a recent change I made to "improve" the reroute nodes behavior :/ I'll look into fixing this. Btw did your problem with the comment boxes go away after changing the Use Knot Node Pool setting?

As for the false branches, is this the layout you mean? Otherwise could you screenshot the BP you have for this one too? I've enabled the "Expand By Height" setting by default which most likely caused the issue you are having.

[TobyTheKiwi]

hi fpwong
I never had "Use Knot Node Pool" on so unfortunately that wasn't the cause of the comment problem.


Funnily enough this morning I hit a great example of the weird false branch behaviour. This screenshot pretty well sums it up.

At a guess it seems like the logic to nicely arrange the nodes from a switch statement are overriding the normal logical blocking. On the right side is the problem I first commented on where the false node is much farther to the bottom right that it needs to be.

I hope this helps.

[Antidamage]

There is sometimes a crash under the following circumstances:

1. Get struct
2. Create break struct node and connect one pin to something
3. Click on break struct node and click "Hide unconnected pins"
4. Format graph

Under 4.24 it would just re-show the unconnected pins, but under 4.25 it sometimes crashes. I wasn't able to replicate it every time, but I've had it happen a few times. I'll try to remember to get a crash dump if it happens again.