05:56 <guso78> J24 so you means an Edge of an enclosed cavity? Can send a Pic or testcase?

05:56 <J24k48> i did put the test code in

05:57 <J24k48> a 90° concave corner is not selectable

05:58 <J24k48> difference(){cube(50);translate([-.1,-1,5])cube([2,500,500]);}

05:58 <J24k48> guso78

06:00 <J24k48> also annoying is that the old measurements are still visible after you changed the geometry and even after a new render - i guess you could compare a new render but there also should be an option to clear them

06:04 <J24k48> or maybe make the button also working in preview to clear

06:06 <J24k48> maybe best to auto clear with preview as that mode is not connected to the measurement and resolution changes in $preview can confuse you with the old markings

06:08 <J24k48> and i think it would be great if a line of the actual measurement is added as i had several cases where i was wondering what distance is now measured if the lines are not parallel

06:09 <J24k48> guso78 https://en.wikipedia.org/wiki/Skew_lines

06:14 <J24k48> as you can see here https://imgur.com/a/T3rtF71  the 45 is only the z distance which is not clear

13:30 <guso78k> J24k48, the measured distance between two lines is always the minimum possible. thus  its direction is  the vector product of both lines.

13:30 <J24k48> i just say this might not be clear to every user

13:31 <guso78k> admittedly it would be more clear to display how  it was  measured, but with GL even the  selected edge itself was (pain-in-the-ass) and  its width is unpredictable

13:31 <guso78k> but yes, good input, will look into it as soon as possible/applicable

13:33 <J24k48> or just put in the text - "minimal distance: ..  " if a line is selected

13:34 <guso78k> this is a really easy fix  :')

13:34 <J24k48> besides minimal distance is not right as you can see in the image it is only the z distance without x/y

13:34 <J24k48> the minimal distance should be the hypotenuse

13:35 <J24k48> while z only might be more useful in most cases

13:35 <guso78k> right now i am facing another ridiculous other error: SizeOf(CubeNode) is 128 bytes normally, but when I compile with JavaScript , CubeNode is only 96 bytes and leads to segfaults ...

13:36 <InPhase> The distances are quite nonsensical in a lot of cases.

13:36 <guso78k> J24k48 yes, it should be really easy to implement many different "modes"  of measurement. maybe display the x/y/z distances separately could also help already

13:37 <InPhase> cube([10,5,2]); cube([6,10,4]); With that model, I select the plane at x=10, and the plane at x=6 with the marking diagonal from (6,0,4) to (6,5,2) and I get... 4.41002? Even after thinking about it for a while, I couldn't figure out what it could possibly be measuring.

13:37 <J24k48> yes that would be a nice feature (xyz dist)

13:38 <InPhase> I expected 4. :)

13:38 <InPhase> Given that 10 - 6 is typically 4.

13:38 <J24k48> at the moment i often crash scad when opening put something in and F6 (without F5) .. and it just closes down

13:40 <InPhase> If it is doing some sort of minimal edge distance from the tesselation, then this is not particularly intuitive.

13:41 <guso78k> yes, measuring works on tessellated data, thus also "internal" edges are selectable ... could also be improved

13:41 <InPhase> If it's limited to minimal edge distances, then it is hard to work with because I cannot seem to select concave edges, so there aren't a lot of points to measure.

13:42 <InPhase> Like the only way I can find to get the value 4 on this is to select (10,0,0) to (10,0,2) and (6,0,2) to (6,0,4), which works only because there is a single point in common.

13:43 <InPhase> But (6,0,2) to (6,5,2) and (6,5,0) to (6,5,2) are both unselectable, which would be the natural locations to measure.

13:43 <InPhase> And obviously those have edges.

13:53 <InPhase> cube([10,5,2]); translate([6,5,0]) rotate([0,0,45]) translate([-10,0,0]) cube([20,6,2]); When I morph it to this, again trying to measure 4, I have nothing I can click except this, which then oddly gives 0: https://imgur.com/a/p2NLXIx

13:55 <guso78k> InPhase, these two lines are actually crossing, thus the minimal distance is 0

13:57 <guso78k> to measure only the "visible"  distance between two edges, we need a more sophisticated distance algorithm

14:03 <InPhase> guso78k: They are crossing if they are lines, but not if line segments.

14:03 <InPhase> Does it extend edges to infinity?

14:04 <guso78k> yes, it  does

14:05 <InPhase> Hmm. Well that's kind of confusing. :)

14:05 <guso78k> yes it is . they are lines rather segments, would need some sort of rework

14:05 <InPhase> If it were to be lines, probably lines should be displayed.

14:06 <InPhase> Then at least it would be obvious what's happening.

14:06 <InPhase> But if I see segments, I expect segments to be measured by some modality.

14:06 <guso78k> the code to calculate distance of 2 lines is quite  small, would be  easy to  improve

14:08 <InPhase> Yeah. I could reinvent a shortest segment distance algorithm, but math majors have thought about this more and posted their work online, so it should be straightforward. :)

14:10 <InPhase> I guess if we had all edges selectable, and true segment distances, along with proper documentation for this explaining it, that would probably fix up most of the issues.

14:12 <InPhase> I hadn't really tried the feature until it came up in discussion here, and wasn't following the development, so my quick test gave the false intuition that the diagonals were a plane selection. I think one of my early tests convinced me that it was a plane selection because of how one of the tesselations just happened to work out and the answer I got. But I consider that misconception fixable with

14:12 <InPhase> documentation.

14:15 <guso78k> sorry, no plane selection yet, only edges and points

14:15 <guso78k> also point to edge distance is  rather a point-to-line distance

22:44 <gbruno> [github] seanmahrt opened issue #5170 (need to clean out old openscad installer and reset back to clean sheet) https://github.com/openscad/openscad/issues/5170