layerling 1.143.0

• GT2 pulley (Discussion #1319): There was no timing belt pulley; a gear's teeth do not fit a belt. A new shape next to the gear and the Hirth joint, pulley, is a pulley for a 2 mm GT2 belt: the groove is the GT2 belt tooth made of true arcs (tip 0.555 mm, 0.75 mm deep, flanks of 1 mm radius, 0.15 mm fillets to the outside circle), and the outside diameter is teeth x 2 / pi less twice the 0.254 mm pitch line differential, 12.22 mm for 20 teeth. Any number of teeth from 10 to 200; from the bottom up a flange, the toothed band (pulleyBeltWidth, default 7 for a 6 mm belt), a flange and a hub, with pulleyFlange (0 for none), pulleyFlangeRise, pulleyHubHeight (0 for none), pulleyHubDiameter and the bore. Width, depth and height follow from these: a side handle or a typed width picks the teeth, the top handle or a typed height the belt width. It is an exact body for the edge tools and STEP (the band pushed up, the flanges and hub joined, the bore cut), its display mesh is closed and drawn from the same arcs. Properties panel with the outside and pitch diameters, saved shape defaults, MCP (layerling_create_shape, layerling_update_object) and skill, guide in all languages, README, start page, card. Asked for by @mrfoameruk.
• Hirth joint with saw teeth, teeth down to the bottom (Discussion #1320): The Hirth joint only had symmetric V teeth, and its tooth height stopped 0.4 mm short of its height, so a 3 mm joint had at most 2.6 mm teeth. A Hirth joint now takes hirthSteepFlank (45 to 90 degrees, missing for symmetric teeth): one flank of every tooth stands at that angle to the face at the rim, 90 degrees being a step, and the other takes the rest of the tooth, like a ratchet; hirthReverse puts the steep flank on the other side. The root then lies a share of a tooth past its crest worked out from the angle, the tooth height and the pitch, and the crest is set so that a copy turned over front to back still locks into it. A step is a wall in the plane through the axis, in the display mesh (kept closed, the outer and bore walls take the step's corner) and in the exact body for the edge tool and STEP. The tooth height now goes up to the joint's full height: the roots touch the bottom at the rim, and the joint still holds together towards the axis. A joint saved with a stored tooth height above its old limit (made lower with the handle, which keeps the stored value) now shows that taller tooth; the properties panel always kept it within the limit. Inspector (tooth shape, steep flank angle, other side, the gentle flank's angle in the hint), saved shape defaults, MCP tools, guide (EN/DE), README, start page and card.
• Surface texture: squares and triangles (Discussion #1302): The closest the textures came to a grid of equal tiles was grooves, which cannot keep the same gap between the lines. Two new patterns lay tiles of one size in a grid, raised or pressed in: squares, and triangles in rows pointing up and down in turn. The spacing is a tile's side and tileGap the gap between two tiles in mm; both stay exactly as set, so 3.2 mm squares 3.2 mm apart are eight lines of a 0.4 mm nozzle, unlike bricks, which are fitted to the edges. The grid is centred on each face that gets it; round a round body a whole number goes round and the gap is rounded a little. Flat tiles start raised, 4 mm (triangles 6 mm) with a 2 mm gap; pointed makes pyramids, round and half round cushions. The walls are cut along the pattern's lines, so every tile stays flat. tileFromBottom ("From the bottom edge") starts the first row right at the bottom edge, so the tiles reach down into the first layer for a brim. Built by @marcus040473 in #1329, asked for by @MichaelDCarney2.
• Enclosure parts are exact bodies (#1327): Screw boss, snap-fit and rib were triangle meshes only, so Fillet, Chamfer, Hollow and the STEP export could not take them. They are now built as exact bodies from the same plans and outlines as their meshes, with true arcs for the hook's fillet and the rib's round top; the mesh stays for the view and as a fallback. A boss set as a hole is its pocket alone, standing where the editor draws it, from its floor up to the boss's top (Discussion #1288). Built by @marcus040473.
• Helical and herringbone gears with more than 30 involute teeth go into STEP (#1321): Above 480 tooth-ring corners, about 30 involute teeth, such a gear got no exact body, for the edge tool and the STEP export alike, so the export left it out and said "Nothing here can be written as STEP". The STEP export now builds it at any number of teeth (cadModifierHelicalGearForShape(shape, { anySize: true })); it runs in the geometry worker, and with the kernel in Node a herringbone gear of 96 teeth builds in 3.7 s and writes in 0.9 s, one of 500 teeth in 36 s and 11 s. The edge tool keeps the limit, so its preview stays inside its timeout. Guide in all languages. Reported by @Elkniwt.
• A part hidden in an open group: the parts left keep their shape (#1323): Hidden while the group was open, a part stayed hidden after Done, but the viewport measured the group's content without it and stretched the parts left to fill the group's box, which keeps the hidden part. The content is now measured with the hidden parts, as the export, the split and the alignment already measured the group's size. The alignment also counts a part hidden inside a group, so a wide hidden cylinder on an axle keeps the group centred on the axle (alignBounds); the export and the group's mesh still leave it out. Guide chapter 4 in all languages. Reported by @Elkniwt.
• Align: equal gaps on stacked parts and how to leave Align say so (Discussion #1332): Parts duplicated on one spot have no room between the outer ones, so the equal-gaps dot was greyed out with "Not enough room" and did not say that a click on the empty plate makes the plate the reference and spreads them over it; a second click there only lets the plate go, which looked like Align would not end. The dot's tooltip now names the click on the plate when the plate is not the reference yet (status.spreadNoRoomPlate), and Align's start message says that L or Esc ends it. All languages. Reported by @makinglayerschannel.
• Tape measure on a cylinder's rim: With "Measure whole edges" on, a click on the top rim of a cylinder set a single point instead of measuring the rim: each side line of the cylinder ends on the rim, about 0.6 mm apart on a 20 mm cylinder, and the nearest end won as a corner over the edge. A half sphere has no side lines there, so it worked. While measuring whole edges, a corner that only sits on a closed round edge within reach now gives way to that edge, and the click gives the diameter. Real corners, the protractor and clicks with whole edges off stay as they were. Asked for by @makinglayerschannel.
• Vase drawing: a double-click on the line adds a point in the browser too (Discussion #1252): The earlier fix waited for a double-click event that a real browser never sends there: letting go of the first click redraws the line, and Chrome then sends neither click nor double-click. The drawing now counts the clicks itself: two clicks on the line without dragging, at most 500 ms and 6 px apart, add the point. Reported by @forrrge.
• Flush lid hinge: no sliver at the hinge's ends (#1326): On a box with rounded corners, the flush hinge left a sliver of the lid's plate 0.01 mm thin at both ends and, in the box, a fin of the wall beside the end knuckles, because the cuts round the axis stopped a gap short of the end knuckle. The plate is now cut on into the room round the end knuckle and the wall right up to it. Reported in the layerling Facebook group, fixed by @marcus040473.
• Long button names in the installed app stay on the screen (Discussion #1333): In the installed app a button's name shows in a small bubble above the button, placed as if it were always one line. Near the right edge the bubble also shrank to the room left and grew tall, so a long name such as the shape lock's ran off the top of the window. The bubble now keeps its full width and works out its place with its real height: when it does not fit above, it goes below the button. Reported by @dlguess.