discuss@lists.openscad.org

OpenSCAD general discussion Mailing-list

View all threads

strange F5 behavior

J
jon
Tue, Jun 27, 2023 9:47 PM

Usually, when I do F5, something happens in the UI to show that the
process has started, even if it takes a few minutes to complete.

When I do F5 with the following code, nothing happens.  The cursor does
not change from an arrow to an hourglass.  Nothing shows up in the
Console.  And this can be quite confusing.

This code completes F5 in about 58 seconds, but the first sign of life
occurs at around 55 seconds.

I installed the latest build of OpenSCAD and BOSL2.

include <BOSL2/std.scad>
inches = 25.4;
sl = 7 * inches;
sw = 2 * inches;
th = 2;
eps = 0.01;

    difference() {
        tex = texture("rough", roughness = 0.5);
        linear_sweep(regular_ngon(n=3, d = sw + 2th), h=sl);
        translate([0, 0, -eps])
            linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl +
2
eps, tex_size=[10,10], style="min_edge");
        }

Usually, when I do F5, something happens in the UI to show that the process has started, even if it takes a few minutes to complete. When I do F5 with the following code, nothing happens.  The cursor does not change from an arrow to an hourglass.  Nothing shows up in the Console.  And this can be quite confusing. This code completes F5 in about 58 seconds, but the first sign of life occurs at around 55 seconds. I installed the latest build of OpenSCAD and BOSL2. include <BOSL2/std.scad> inches = 25.4; sl = 7 * inches; sw = 2 * inches; th = 2; eps = 0.01;     difference() {         tex = texture("rough", roughness = 0.5);         linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl);         translate([0, 0, -eps])             linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl + 2*eps, tex_size=[10,10], style="min_edge");         }
AM
Adrian Mariano
Tue, Jun 27, 2023 9:52 PM

I get "Loaded design..." when the preview computation starts, and then not
much until it finishes, which is after 10s for me.  I think it's pretty
normal that "loaded design" is all I see to indicate a computation has
started.

On Tue, Jun 27, 2023 at 5:48 PM jon jon@jonbondy.com wrote:

Usually, when I do F5, something happens in the UI to show that the
process has started, even if it takes a few minutes to complete.

When I do F5 with the following code, nothing happens.  The cursor does
not change from an arrow to an hourglass.  Nothing shows up in the
Console.  And this can be quite confusing.

This code completes F5 in about 58 seconds, but the first sign of life
occurs at around 55 seconds.

I installed the latest build of OpenSCAD and BOSL2.

include <BOSL2/std.scad>
inches = 25.4;
sl = 7 * inches;
sw = 2 * inches;
th = 2;
eps = 0.01;

  difference() {
      tex = texture("rough", roughness = 0.5);
      linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl);
      translate([0, 0, -eps])
          linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl +

2*eps, tex_size=[10,10], style="min_edge");
}


OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

I get "Loaded design..." when the preview computation starts, and then not much until it finishes, which is after 10s for me. I think it's pretty normal that "loaded design" is all I see to indicate a computation has started. On Tue, Jun 27, 2023 at 5:48 PM jon <jon@jonbondy.com> wrote: > Usually, when I do F5, something happens in the UI to show that the > process has started, even if it takes a few minutes to complete. > > When I do F5 with the following code, nothing happens. The cursor does > not change from an arrow to an hourglass. Nothing shows up in the > Console. And this can be quite confusing. > > This code completes F5 in about 58 seconds, but the first sign of life > occurs at around 55 seconds. > > I installed the latest build of OpenSCAD and BOSL2. > > > include <BOSL2/std.scad> > inches = 25.4; > sl = 7 * inches; > sw = 2 * inches; > th = 2; > eps = 0.01; > > difference() { > tex = texture("rough", roughness = 0.5); > linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl); > translate([0, 0, -eps]) > linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl + > 2*eps, tex_size=[10,10], style="min_edge"); > } > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org >
J
jon
Tue, Jun 27, 2023 10:15 PM

Adrian:

I wonder how you can get that to finish in 10 seconds when I am running
on a Windows machine with 4 GHz processors.  Puzzling.

I have been developing a 3D design for a few days, and only added the
code snippet when I discovered the "rough" texture (thank you very much,
BOSL2)!  I can tell you that my development has slowed down to a crawl. 
F5s that took under a second, now take 3-4 minutes.  And nothing happens
at all until the last few seconds.

I added a render() just before the difference(), hoping that the
structure might be computed once and not re-calculated each time I hit
F5, but no joy at all.

Any thoughts about how to speed optimize this?

Thanks!

Jon

On 6/27/2023 5:52 PM, Adrian Mariano wrote:

I get "Loaded design..." when the preview computation starts, and then
not much until it finishes, which is after 10s for me.  I think it's
pretty normal that "loaded design" is all I see to indicate a
computation has started.

On Tue, Jun 27, 2023 at 5:48 PM jon jon@jonbondy.com wrote:

 Usually, when I do F5, something happens in the UI to show that the
 process has started, even if it takes a few minutes to complete.

 When I do F5 with the following code, nothing happens.  The cursor
 does
 not change from an arrow to an hourglass.  Nothing shows up in the
 Console.  And this can be quite confusing.

 This code completes F5 in about 58 seconds, but the first sign of
 life
 occurs at around 55 seconds.

 I installed the latest build of OpenSCAD and BOSL2.


 include <BOSL2/std.scad>
 inches = 25.4;
 sl = 7 * inches;
 sw = 2 * inches;
 th = 2;
 eps = 0.01;

      difference() {
          tex = texture("rough", roughness = 0.5);
          linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl);
          translate([0, 0, -eps])
              linear_sweep(regular_ngon(n=3, d = sw), texture=tex,
 h=sl +
 2*eps, tex_size=[10,10], style="min_edge");
          }
 _______________________________________________
 OpenSCAD mailing list
 To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

Adrian: I wonder how you can get that to finish in 10 seconds when I am running on a Windows machine with 4 GHz processors.  Puzzling. I have been developing a 3D design for a few days, and only added the code snippet when I discovered the "rough" texture (thank you very much, BOSL2)!  I can tell you that my development has slowed down to a crawl.  F5s that took under a second, now take 3-4 minutes.  And nothing happens at all until the last few seconds. I added a render() just before the difference(), hoping that the structure might be computed once and not re-calculated each time I hit F5, but no joy at all. Any thoughts about how to speed optimize this? Thanks! Jon On 6/27/2023 5:52 PM, Adrian Mariano wrote: > I get "Loaded design..." when the preview computation starts, and then > not much until it finishes, which is after 10s for me.  I think it's > pretty normal that "loaded design" is all I see to indicate a > computation has started. > > On Tue, Jun 27, 2023 at 5:48 PM jon <jon@jonbondy.com> wrote: > > Usually, when I do F5, something happens in the UI to show that the > process has started, even if it takes a few minutes to complete. > > When I do F5 with the following code, nothing happens.  The cursor > does > not change from an arrow to an hourglass.  Nothing shows up in the > Console.  And this can be quite confusing. > > This code completes F5 in about 58 seconds, but the first sign of > life > occurs at around 55 seconds. > > I installed the latest build of OpenSCAD and BOSL2. > > > include <BOSL2/std.scad> > inches = 25.4; > sl = 7 * inches; > sw = 2 * inches; > th = 2; > eps = 0.01; > >      difference() { >          tex = texture("rough", roughness = 0.5); >          linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl); >          translate([0, 0, -eps]) >              linear_sweep(regular_ngon(n=3, d = sw), texture=tex, > h=sl + > 2*eps, tex_size=[10,10], style="min_edge"); >          } > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org > > > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email todiscuss-leave@lists.openscad.org
J
jon
Tue, Jun 27, 2023 10:21 PM

BTW, the "Loaded design" message means that the file was loaded. It does
not occur at F5, but before.  The first thing that appears on the
Console when you press F5 is "Parsing design (AST generation)"

Jon

On 6/27/2023 5:52 PM, Adrian Mariano wrote:

I get "Loaded design..." when the preview computation starts, and then
not much until it finishes, which is after 10s for me.  I think it's
pretty normal that "loaded design" is all I see to indicate a
computation has started.

On Tue, Jun 27, 2023 at 5:48 PM jon jon@jonbondy.com wrote:

 Usually, when I do F5, something happens in the UI to show that the
 process has started, even if it takes a few minutes to complete.

 When I do F5 with the following code, nothing happens.  The cursor
 does
 not change from an arrow to an hourglass.  Nothing shows up in the
 Console.  And this can be quite confusing.

 This code completes F5 in about 58 seconds, but the first sign of
 life
 occurs at around 55 seconds.

 I installed the latest build of OpenSCAD and BOSL2.


 include <BOSL2/std.scad>
 inches = 25.4;
 sl = 7 * inches;
 sw = 2 * inches;
 th = 2;
 eps = 0.01;

      difference() {
          tex = texture("rough", roughness = 0.5);
          linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl);
          translate([0, 0, -eps])
              linear_sweep(regular_ngon(n=3, d = sw), texture=tex,
 h=sl +
 2*eps, tex_size=[10,10], style="min_edge");
          }
 _______________________________________________
 OpenSCAD mailing list
 To unsubscribe send an email to discuss-leave@lists.openscad.org

OpenSCAD mailing list
To unsubscribe send an email todiscuss-leave@lists.openscad.org

BTW, the "Loaded design" message means that the file was loaded. It does not occur at F5, but before.  The first thing that appears on the Console when you press F5 is "Parsing design (AST generation)" Jon On 6/27/2023 5:52 PM, Adrian Mariano wrote: > I get "Loaded design..." when the preview computation starts, and then > not much until it finishes, which is after 10s for me.  I think it's > pretty normal that "loaded design" is all I see to indicate a > computation has started. > > On Tue, Jun 27, 2023 at 5:48 PM jon <jon@jonbondy.com> wrote: > > Usually, when I do F5, something happens in the UI to show that the > process has started, even if it takes a few minutes to complete. > > When I do F5 with the following code, nothing happens.  The cursor > does > not change from an arrow to an hourglass.  Nothing shows up in the > Console.  And this can be quite confusing. > > This code completes F5 in about 58 seconds, but the first sign of > life > occurs at around 55 seconds. > > I installed the latest build of OpenSCAD and BOSL2. > > > include <BOSL2/std.scad> > inches = 25.4; > sl = 7 * inches; > sw = 2 * inches; > th = 2; > eps = 0.01; > >      difference() { >          tex = texture("rough", roughness = 0.5); >          linear_sweep(regular_ngon(n=3, d = sw + 2*th), h=sl); >          translate([0, 0, -eps]) >              linear_sweep(regular_ngon(n=3, d = sw), texture=tex, > h=sl + > 2*eps, tex_size=[10,10], style="min_edge"); >          } > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org > > > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email todiscuss-leave@lists.openscad.org
JB
Jordan Brown
Wed, Jun 28, 2023 2:02 AM

Something has changed with respect to how stuff is displayed on the
console.  The output used to be displayed incrementally, as it happened,
and now it is all displayed on completion.

for (k=[1:10]) {
    for (i=[1:1000], j=[1:1000]) {}
    echo(k);
}

On 2021.01, that displays the numbers 1 through 10, about two per
second, for a total of 6 seconds.

On 2023.06.06, it pauses for about 10 seconds and displays all of it,
for a reported total of 10 seconds.

The slowdown is also of interest.

Something has changed with respect to how stuff is displayed on the console.  The output used to be displayed incrementally, as it happened, and now it is all displayed on completion. for (k=[1:10]) { for (i=[1:1000], j=[1:1000]) {} echo(k); } On 2021.01, that displays the numbers 1 through 10, about two per second, for a total of 6 seconds. On 2023.06.06, it pauses for about 10 seconds and displays all of it, for a reported total of 10 seconds. The slowdown is also of interest.
J
jon
Wed, Jun 28, 2023 2:12 AM

I now have some ... simple? ... code that behaves erratically. I've
tried clearing the cache without making any headway.  I am using the
most recent snapshot and BOSL2.  At one point I thought it was the
render(), which I removed and achieved success, but the next time it
failed, so that was just random.

Perhaps the rough internal surface is sometimes generated so that the
difference() works and sometimes it fails.  I do  not know whether the
"rough" texture is computed differently each time, or is static; the
former could explain what I am seeing.

Any other thoughts?  I have no idea whether this is reproducible or not

Jon

// when I F6 this code, sometimes I get
// - a solid bar with a triangular cross section
// - a hollow bar with a triangular cross section and a "rough" interior
finish
// - the error message "ERROR: The given mesh is not closed! Unable to
convert to CGAL_Nef_Polyhedron."

include <BOSL2/std.scad>

inches = 25.4;
eps = 0.01;
th = 2;

sl = 7 * inches;
sw = 1.8 * inches;
//    render()
        difference() {
            tex = texture("rough", roughness = 0.5);
            linear_sweep(regular_ngon(n=3, d = sw + 4*th), h=sl,
convexity = 10);
            translate([0, 0, -1])
// Pick a smooth interior (next line) or a rough interior (line after
that) by commenting one line out
//                linear_sweep(regular_ngon(n=3, d = sw),             
h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);
                linear_sweep(regular_ngon(n=3, d = sw), texture=tex,
h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);
            }

On 6/27/2023 10:02 PM, Jordan Brown wrote:

Something has changed with respect to how stuff is displayed on the
console.  The output used to be displayed incrementally, as it
happened, and now it is all displayed on completion.

 for (k=[1:10]) {
      for (i=[1:1000], j=[1:1000]) {}
      echo(k);
 }

On 2021.01, that displays the numbers 1 through 10, about two per
second, for a total of 6 seconds.

On 2023.06.06, it pauses for about 10 seconds and displays all of it,
for a reported total of 10 seconds.

The slowdown is also of interest.

I now have some ... simple? ... code that behaves erratically. I've tried clearing the cache without making any headway.  I am using the most recent snapshot and BOSL2.  At one point I thought it was the render(), which I removed and achieved success, but the next time it failed, so that was just random. Perhaps the rough internal surface is sometimes generated so that the difference() works and sometimes it fails.  I do  not know whether the "rough" texture is computed differently each time, or is static; the former could explain what I am seeing. Any other thoughts?  I have no idea whether this is reproducible or not Jon // when I F6 this code, sometimes I get // - a solid bar with a triangular cross section // - a hollow bar with a triangular cross section and a "rough" interior finish // - the error message "ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron." include <BOSL2/std.scad> inches = 25.4; eps = 0.01; th = 2; sl = 7 * inches; sw = 1.8 * inches; //    render()         difference() {             tex = texture("rough", roughness = 0.5);             linear_sweep(regular_ngon(n=3, d = sw + 4*th), h=sl, convexity = 10);             translate([0, 0, -1]) // Pick a smooth interior (next line) or a rough interior (line after that) by commenting one line out //                linear_sweep(regular_ngon(n=3, d = sw),              h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);                 linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);             } On 6/27/2023 10:02 PM, Jordan Brown wrote: > Something has changed with respect to how stuff is displayed on the > console.  The output used to be displayed incrementally, as it > happened, and now it is all displayed on completion. > > for (k=[1:10]) { > for (i=[1:1000], j=[1:1000]) {} > echo(k); > } > > On 2021.01, that displays the numbers 1 through 10, about two per > second, for a total of 6 seconds. > > On 2023.06.06, it pauses for about 10 seconds and displays all of it, > for a reported total of 10 seconds. > > The slowdown is also of interest. > >
JB
Jordan Brown
Wed, Jun 28, 2023 2:20 AM

On 6/27/2023 7:12 PM, jon wrote:

I now have some ... simple? ... code that behaves erratically.  I've
tried clearing the cache without making any headway.  I am using the
most recent snapshot and BOSL2.  At one point I thought it was the
render(), which I removed and achieved success, but the next time it
failed, so that was just random.

Perhaps the rough internal surface is sometimes generated so that the
difference() works and sometimes it fails.  I do  not know whether the
"rough" texture is computed differently each time, or is static; the
former could explain what I am seeing.

Any other thoughts?  I have no idea whether this is reproducible or not

I don't have anything useful to say about that; it seems like a Revar /
Adrian question.

But it seems separate from the issue where you don't get incremental
output as the program runs.

On 6/27/2023 7:12 PM, jon wrote: > > I now have some ... simple? ... code that behaves erratically.  I've > tried clearing the cache without making any headway.  I am using the > most recent snapshot and BOSL2.  At one point I thought it was the > render(), which I removed and achieved success, but the next time it > failed, so that was just random. > > Perhaps the rough internal surface is sometimes generated so that the > difference() works and sometimes it fails.  I do  not know whether the > "rough" texture is computed differently each time, or is static; the > former could explain what I am seeing. > > Any other thoughts?  I have no idea whether this is reproducible or not > I don't have anything useful to say about that; it seems like a Revar / Adrian question. But it seems separate from the issue where you don't get incremental output as the program runs.
RD
Revar Desmera
Wed, Jun 28, 2023 2:31 AM

On Jun 27, 2023, at 3:15 PM, jon jon@jonbondy.com wrote:
I have been developing a 3D design for a few days, and only added the code snippet when I discovered the "rough" texture (thank you very much, BOSL2)!  I can tell you that my development has slowed down to a crawl.  F5s that took under a second, now take 3-4 minutes.  And nothing happens at all until the last few seconds.

I added a render() just before the difference(), hoping that the structure might be computed once and not re-calculated each time I hit F5, but no joy at all.

Any thoughts about how to speed optimize this?

Several things of note here.

  1. You can avoid using textures during preview, if you do something like:

tex = $preview? [[0]] : texture("rough", roughness=0.5);

  1. On my laptop an F-5 of this code takes just 7.7s.  Are you running out of memory, maybe?

  2. This code appears to expose a BOSL2 texturing bug where it's using reversed faces.  This probably won't work until we can fix this.

> On Jun 27, 2023, at 3:15 PM, jon <jon@jonbondy.com> wrote: > I have been developing a 3D design for a few days, and only added the code snippet when I discovered the "rough" texture (thank you very much, BOSL2)! I can tell you that my development has slowed down to a crawl. F5s that took under a second, now take 3-4 minutes. And nothing happens at all until the last few seconds. > > I added a render() just before the difference(), hoping that the structure might be computed once and not re-calculated each time I hit F5, but no joy at all. > > Any thoughts about how to speed optimize this? > Several things of note here. 1. You can avoid using textures during preview, if you do something like: tex = $preview? [[0]] : texture("rough", roughness=0.5); 2. On my laptop an F-5 of this code takes just 7.7s. Are you running out of memory, maybe? 3. This code appears to expose a BOSL2 texturing bug where it's using reversed faces. This probably won't work until we can fix this.
RD
Revar Desmera
Wed, Jun 28, 2023 2:35 AM

Actually, even faster texture-less previews can be done with:

tex = $preview? undef : texture("rough", roughness=0.5);

-Revar

On Jun 27, 2023, at 7:31 PM, Revar Desmera revarbat@gmail.com wrote:

On Jun 27, 2023, at 3:15 PM, jon jon@jonbondy.com wrote:
I have been developing a 3D design for a few days, and only added the code snippet when I discovered the "rough" texture (thank you very much, BOSL2)!  I can tell you that my development has slowed down to a crawl.  F5s that took under a second, now take 3-4 minutes.  And nothing happens at all until the last few seconds.

I added a render() just before the difference(), hoping that the structure might be computed once and not re-calculated each time I hit F5, but no joy at all.

Any thoughts about how to speed optimize this?

Several things of note here.

  1. You can avoid using textures during preview, if you do something like:

tex = $preview? [[0]] : texture("rough", roughness=0.5);

  1. On my laptop an F-5 of this code takes just 7.7s.  Are you running out of memory, maybe?

  2. This code appears to expose a BOSL2 texturing bug where it's using reversed faces.  This probably won't work until we can fix this.

Actually, even faster texture-less previews can be done with: tex = $preview? undef : texture("rough", roughness=0.5); -Revar > On Jun 27, 2023, at 7:31 PM, Revar Desmera <revarbat@gmail.com> wrote: > > > >> On Jun 27, 2023, at 3:15 PM, jon <jon@jonbondy.com> wrote: >> I have been developing a 3D design for a few days, and only added the code snippet when I discovered the "rough" texture (thank you very much, BOSL2)! I can tell you that my development has slowed down to a crawl. F5s that took under a second, now take 3-4 minutes. And nothing happens at all until the last few seconds. >> >> I added a render() just before the difference(), hoping that the structure might be computed once and not re-calculated each time I hit F5, but no joy at all. >> >> Any thoughts about how to speed optimize this? >> > > Several things of note here. > > 1. You can avoid using textures during preview, if you do something like: > > tex = $preview? [[0]] : texture("rough", roughness=0.5); > > 2. On my laptop an F-5 of this code takes just 7.7s. Are you running out of memory, maybe? > > 3. This code appears to expose a BOSL2 texturing bug where it's using reversed faces. This probably won't work until we can fix this. >
RD
Revar Desmera
Wed, Jun 28, 2023 4:15 AM

Texture face reversal bug has just been fixed.  Please download the latest BOSL2 code.

- Revar

On Jun 27, 2023, at 7:12 PM, jon jon@jonbondy.com wrote:

I now have some ... simple? ... code that behaves erratically.  I've tried clearing the cache without making any headway.  I am using the most recent snapshot and BOSL2.  At one point I thought it was the render(), which I removed and achieved success, but the next time it failed, so that was just random.

Perhaps the rough internal surface is sometimes generated so that the difference() works and sometimes it fails.  I do  not know whether the "rough" texture is computed differently each time, or is static; the former could explain what I am seeing.

Any other thoughts?  I have no idea whether this is reproducible or not

Jon

// when I F6 this code, sometimes I get
// - a solid bar with a triangular cross section
// - a hollow bar with a triangular cross section and a "rough" interior finish
// - the error message "ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron."

include <BOSL2/std.scad>

inches = 25.4;
eps = 0.01;
th = 2;

sl = 7 * inches;
sw = 1.8 * inches;
//    render()
difference() {
tex = texture("rough", roughness = 0.5);
linear_sweep(regular_ngon(n=3, d = sw + 4*th), h=sl, convexity = 10);
translate([0, 0, -1])
// Pick a smooth interior (next line) or a rough interior (line after that) by commenting one line out
//                linear_sweep(regular_ngon(n=3, d = sw),              h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);
linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10);
}

On 6/27/2023 10:02 PM, Jordan Brown wrote:

Something has changed with respect to how stuff is displayed on the console.  The output used to be displayed incrementally, as it happened, and now it is all displayed on completion.
for (k=[1:10]) {
for (i=[1:1000], j=[1:1000]) {}
echo(k);
}
On 2021.01, that displays the numbers 1 through 10, about two per second, for a total of 6 seconds.

On 2023.06.06, it pauses for about 10 seconds and displays all of it, for a reported total of 10 seconds.

The slowdown is also of interest.


OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org

Texture face reversal bug has just been fixed. Please download the latest BOSL2 code. - Revar > On Jun 27, 2023, at 7:12 PM, jon <jon@jonbondy.com> wrote: > > I now have some ... simple? ... code that behaves erratically. I've tried clearing the cache without making any headway. I am using the most recent snapshot and BOSL2. At one point I thought it was the render(), which I removed and achieved success, but the next time it failed, so that was just random. > > Perhaps the rough internal surface is sometimes generated so that the difference() works and sometimes it fails. I do not know whether the "rough" texture is computed differently each time, or is static; the former could explain what I am seeing. > > Any other thoughts? I have no idea whether this is reproducible or not > > Jon > > > > // when I F6 this code, sometimes I get > // - a solid bar with a triangular cross section > // - a hollow bar with a triangular cross section and a "rough" interior finish > // - the error message "ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron." > > include <BOSL2/std.scad> > > inches = 25.4; > eps = 0.01; > th = 2; > > sl = 7 * inches; > sw = 1.8 * inches; > // render() > difference() { > tex = texture("rough", roughness = 0.5); > linear_sweep(regular_ngon(n=3, d = sw + 4*th), h=sl, convexity = 10); > translate([0, 0, -1]) > // Pick a smooth interior (next line) or a rough interior (line after that) by commenting one line out > // linear_sweep(regular_ngon(n=3, d = sw), h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10); > linear_sweep(regular_ngon(n=3, d = sw), texture=tex, h=sl + 2, tex_size=[10,10], style="min_edge", convexity = 10); > } > > > > > On 6/27/2023 10:02 PM, Jordan Brown wrote: >> Something has changed with respect to how stuff is displayed on the console. The output used to be displayed incrementally, as it happened, and now it is all displayed on completion. >> for (k=[1:10]) { >> for (i=[1:1000], j=[1:1000]) {} >> echo(k); >> } >> On 2021.01, that displays the numbers 1 through 10, about two per second, for a total of 6 seconds. >> >> On 2023.06.06, it pauses for about 10 seconds and displays all of it, for a reported total of 10 seconds. >> >> The slowdown is also of interest. >> >> >> > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org