whitequark changed the topic of #solvespace to: SolveSpace--parametric 2d/3d CAD · latest version 2.3 · http://solvespace.com · code at https://github.com/solvespace/solvespace · logs at https://irclog.whitequark.org/solvespace
<wpwrak> as, something that would be nice is if center points of arcs/circles (and similar "special" points) would somehow be indicated as such in TW. something like s/POINT/CENTER POINT/, a centter-of-arc constraint, or whatever makes sense
<whitequark> hm
<whitequark> that would be fairly trivial to do but I'm too lazy to do it
<whitequark> file a bug I guess
<wpwrak> gladly :)
<GitHub> [solvespace] wpwrak opened issue #165: make arc-center relation "explorable" https://github.com/solvespace/solvespace/issues/165
<wpwrak> meh. length difference not supported for arc-to-circle. tz tz.
<whitequark> wpwrak: there isn't a goal to support every permutation of every constraint
<whitequark> if it can be easily done with construction and isn't a super common case, it's probably not worth the dev time
<wpwrak> a lot can be done with construction :)
<whitequark> yup, that's the power of it
<whitequark> implementing an additional constraint is fairly heavyweight
<wpwrak> ah, i was hoping it would be straightfoward
<whitequark> you have to implement it and test that it saves correctly, displays in the UI, etc, etc
<whitequark> then you break compatibility with old versions
<whitequark> for now, all the files you save are openable in 2.3
<whitequark> then you have to support this constraint type, forever
<whitequark> there is already a *lot* and many refactorings have to touch all of them
<wpwrak> yeah, file compatibility is suckish
<whitequark> some day you could define constraints in lua scripts
<wpwrak> that would be handy
<GitHub> [solvespace] wpwrak commented on issue #165: I forget an even more common case: the endpoints of lines. https://github.com/solvespace/solvespace/issues/165#issuecomment-274355963
<wpwrak> hah, got it hang: export a screenshot, then "glGetError() == 0x502" followed by "(solvespace:26896): Gtk-WARNING **: fb setup not supported" and hang
<wpwrak> let's see if i can reproduce it ..
mumptai has joined #solvespace
<wpwrak> but what i actually wanted to do it show this: http://downloads.qi-hardware.com/people/werner/slvs/an-side.png
<wpwrak> pretty neat work of the solver there. scripting this with freecad would have taken a while, and i wouldn't have had nearly as good an idea of how the dimensions affect each other.
<wpwrak> minor nit: the exported PNG doesn't really need to say "previewing exported geometry", does it ? also, is it really "_pre_viewing" ?
<wpwrak> ah, the "previewing" bit seems to be more complicated. i guess it came from the 2D export (to PDF) i did right before, not from the screenshot
<wpwrak> in any case, screenshots (export image) do odd things. this is what my screen looks like after opening a new part, then exporting: http://downloads.qi-hardware.com/people/werner/slvs/new-export-image.png
<wpwrak> the menu bar still works. so i guess it's clearing the canvas of things not to include in the PNG, but the fails to restore them. same diagnostics as above, "glGetError() == 0x502" and "fb setup not supported"
<whitequark> wpwrak: (previewing) so you have render chord tolerance and export chord tolerance
<whitequark> render chord tolerance is in % of bounding box, and export chord tolerance is in millimeters absolute
<whitequark> once you render and before you press Esc you can see the model in the export chord tolerance, since it's already rebuilt
<whitequark> wpwrak: (screenshot) hmm, unable to reproduce
<whitequark> that said... the screenshot code is *horrible*
<whitequark> I was gonna fix it for some time, I guess now is as good as ever?
<wpwrak> (preview) what i mean is, at the time the "preview" is shown, it seems that the file is already written. so it's not really a _pre_view, more like a postview :)
<wpwrak> (reproduce) go you get the glGetError() == 0x502 ?
<whitequark> (preview) oh yeah, kinda
<whitequark> (reproduce) I do
<wpwrak> the exact sequence is as follows: 1) start slvs, 2) File > Expore Image, 3) accept, 4) move the mouse a little on the canvas. then the menu bar may or may not disappear, and the icon block so far has always disappeared
mumptai has quit [Remote host closed the connection]
<whitequark> *sigh* I have no idea what goes wrong in your case nor can I reproduce
<whitequark> but it also doesn't matter
<wpwrak> ah, the glGetError and one "fb setup not supported" appear immediately after saving, when the screen is still intact. then, when i move the mouse, things disappear and more "fb setup not supported" are shown
<wpwrak> all this looks as if the code did something like detach_from_fb(canvas); redraw(canvas); save(cavas); reattach_to_fb(canvas); /* fails */ return; /* to event loop */
<wpwrak> and then the next update tries use the mis-attached canvas
<whitequark> that's ... hm
<whitequark> ah yes I can reproduce
<whitequark> w/e I'm not going to let the hack used for screenshots survive for any longer
<wpwrak> heh :)
<wpwrak> hmm, there may be a potential trap when replicating arcs. scenario: i have a sketch with arcs. now i want to reuse it in another sketch. to "replicate" the arcs, i have to draw them again. the endpoints are easy enough. now the diameter. since there is no diameter equality constraint, i need to find somethine else. if i have a third point on the original arc, great. else, i could use the center point. however, if i use "O" for the
<wpwrak> centers, i try to kill one DOF too many.
<wpwrak> if i use H or V, i get what i want. however, now the whole thing should get undefined at certain angles.