Better guessing at input values


The inference system is great, but if I am drawing a rectangle (already placed the starting point) and get one dimension to snap/drag along an inference line, I would like to just type in a single value and have it always assume that it’s the dimension not inferred.

If I drag to create a rectangle and type in xxx,yyy I would like it to assume that the larger number I type matches the longest side I have dragged.
(The number of times I have had to re-type to flip the dims is staggering… perhaps a special character could be used in the input box to say “reverse the last input dimensions” eg just type in “/ [return]” and the previous “4,9” would be re-input as “9,4” Or hitting [tab] would insert the last typed value in the dimensions box and hitting it again would rotate the values between separators- hitting [return] would accept these values as if typed in.)

If I infer a single point before the first click and move the mouse away from that point, keeping on the inference line, I would like to type in a value followed by [return] and have the insertion point (first click) placed at this distance from the inference starting point. Alternatively a guideline or inference node be placed at this point.

I’m sure I’ve mentioned this before, but I would like to type in a prefix (eg “+”) to draw another line in the same direction as the last one, with the length typed in.
(eg type “1.3 [return]” draws a line from the start click in the direction of mouse travel by this dimension. If I want to add another line in the same direction I have to currently move the mouse to infer in the same direction and type in the second dimension. If I could type 1.3 [return] +2.4 [return] +0.5 [return] + 1.4 [return] and have it draw 4 lines it would save moving my hand back and fore from keypad to mouse to keypad to mouse…)

In using other CAD packages, any text entry of dimensions pops up a floating input box beside the mouse cursor - might be nice if SU could do the same. Same with the ‘entity info’ - if it was accessed via r-click menu, or better yet if r-clicking had this info displayed (editable) at the top of the menu. Maybe dragging a value from this could insert a text object attached to the entity.

If a text box contained information that matched the ‘entity info’ of the tagged entity, then an “update” button displayed when editing the text would change the entity properties to whatever was typed in (in the same way as it would if changed in the ‘entity info’ box)

I would also like to see text entities to automatically be placed on a specific layer (specified in ‘options’) when they are generated. They could be subsequently moved.


You don’t mention the existing capability to skip an entry field by typing a comma (or semi-colon if required by your locale). For example, “,3” means “keep the first inferred value and change the second one to 3”. Similarly, “6,” overrides the first while keeping the second. In my mind this omission raises the question: were you unaware of this behavior, or are you aware of it and don’t like it?

Your thoughts?



I believe this would be confusing. Normally text callouts and dimension text, etc., use “autotext” tags that reference a property of some object. The simplest is the “<>” autotext which references the dimension itself’s distance property.

It would probably be error prone to have unconventional or varying reference directions in the code. The callouts or dimensions need to be able to reference stable data. That means keeping it in one place.

Adding another place to change an object’s properties will just confuse the userbase and complicate maintaining the User Guide.

What we need is for SketchUp to catch up to Layout with regard to autotext tags.


I know this has been requested before, and likely exists as a plugin.
If I had to guess, I’d say TIG has one. (I know he did one to automatically move primitives to “Layer0”.)


Why nice? Is the on-screen editbox not readable ? (ie, too small?)

Then it cannot be a menu. A menu is a list of links. For it to have edit controls it would be a dialog box. Since the EntityInfo is already a dialog,… there is no point, other than having it appear near to where the cursor is. (I do not know any CAD application that does this. All those I have seen have the property sheet in a dialog window of docked in a window tray.)



Don’t know if I was unaware or forgot about it, but it still doesn’t eliminate the dyslexic problem of mixing up the x and y… I suppose I could just look at the coordinates before I type? hmmm… am I too lazy to move my eyeballs?

[quote=“DanRathbun, post:3, topic:15024”]
I believe this would be confusing. Normally text callouts and dimension text, etc., use “autotext” tags that reference a property of some object. The simplest is the “<>” autotext which references the dimension itself’s distance property.[/quote]
I agree that the autotext is the way to go for dynamic labeling - (obviously only show the control codes when editing) but wouldn’t it be handy to over-write “<x,y>” with “<888,y>” and have the x coordinate change?
If you are dimensioning something and see an error, it could quickly be changed: I think that if you see a dimension displayed anywhere, you should be able to change it and the object it’s relating to changes to match - I can’t concede the point that it would be more confusing to users.

[quote=“DanRathbun, post:4, topic:15024”]I know this has been requested before, and likely exists as a plugin.
[/quote] It doesn’t even need a plugin; one click and I have a new layer, one click and all the new geometry (ie dimensions) gets added to it. The point isn’t that it can already be done, the point is that it would be nice to have the program intuitively do it automatically.

Re floating input boxes -
I think that Maya also has a really cool and intuitive context sensitive menu system that pops up around the mouse.

I know that there is no current control that has this info in it - that’s why it’s a feature request; combining the r-click menu with the entity info box.


You miss my point. On the Windows platform a menu can only have menuitems. There is no class of window (that I know of) that combines the two (dialog controls and menuitems.)

If you look at the current inspector dialogs, you’ll see a special button that displays a menu. The opposite could also be done, the context menu could have an item that displays a dialog box.

It is unlikley that the Trimble Team would invent a completely new UI window class. (At least in the short term.)

But sure it might be nifty to have a mouse-following property sheet like ArchiCAD’s. It could likely be done with a custom select tool.


No because “<888,y>” would be an invalid autotext tag.
You’d have to type “888,<y>” but that would override whatever was displayed for the x value, from then on.

So no, I am not at all for changing the long established protocol for overriding text objects and shoe-horning some property setting function into it.

Hey I have 37 years at this, and that is just plain butt-backward.

Dimensions are associated WITH objects. Each end of the dimension’s extension lines can be anchored to a different object. It is moving or stretching these objects that causes an associative dimension to automatically update it’s text (depending upon the format of the plain text and autotext tag in the text field.)

Now you want to turn everything upside down, … ie, the way users work with dimensions ? … the way the code deals with dimensions ?

Sorry … :-1: :-1:


you only have a couple of years on me, but I have worked with several custom drawing packages and tools designed for salesmen to use where you have to really dumb things down and simplify them. If there is a dimension label - why can’t you change it and have the object’s dimension change?

You’ve been around long enough to know that just because that’s the way it has been done in the past does not mean that it’s the best way or simplest way - how often do you have to explain to people used to working in other software that you can do things differently?

(PS I was thinking that after over-writing the x to “<888,y>”, hitting [return] it would change the object and close the edit, reverting back to displaying “888,555”. Editing it again you would see “<x,y>”)


Because it is very difficult to program that in code. It is backwards to how it works now. And it just will never happen. So there is no sense wasting any more time arguing about it. (Besides it would be simpler to grab the vertex where both the geometry and the dimension are anchored and stretch them at the same time.)

New features are one thing, little improvements are great, but the SketchUp interface is the way it is BY DESIGN. It is purposely NOT like other CAD programs.

SketchUp uses a GUI. The G in GUI is “graphical”. You are promoting a TUI. A Textual User Interface. I have been there! I was around back in the day when CAD first came on the scene. I remember some CAD programs where entirely text input, before mice became affordable.

I have no desire to go back to the “dark ages” :wink:


Assuming that “X” needs to be an exact value, you will have to type that value in somewhere - rather than going back to the dark ages and having to type in the commands in one place, I am proposing to bring the light and be able to type it in anywhere. Put it this way: is there any other place you can find x where you cannot change it?

The other notable thing is that this proposal assumes dynamic tags - the ability to put every variable that you have available in a dynamic component between tags. So instead of simply “<x>” you could have a label that says “instances : <instances>” changing the word for a numeric value would update the DC as if it had been input in the DC’s dialogue.