discuss@lists.openscad.org

OpenSCAD general discussion Mailing-list

View all threads

debugging a polyhedron....

JB
Jordan Brown
Thu, Jan 20, 2022 4:13 PM

On 1/20/2022 4:03 AM, nop head wrote:

That gives me this, which is clearly broken, even in the preview. Is
that what you see?

image.png

Places where from some angles you can see through the model are a dead
giveaway for convexity problems.  Setting convexity=2 in the
polyhedron() call fixes this.

I still think that we should set the default convexity to 10, to make
the vast majority of these problems go away with only a very small
performance hit.

On 1/20/2022 4:03 AM, nop head wrote: > That gives me this, which is clearly broken, even in the preview. Is > that what you see? > > image.png > Places where from some angles you can see through the model are a dead giveaway for convexity problems.  Setting convexity=2 in the polyhedron() call fixes this. I still think that we should set the default convexity to 10, to make the vast majority of these problems go away with only a very small performance hit.
J
jon
Thu, Jan 20, 2022 4:17 PM

I vote for a default convexity of 100.  We all waste far too much time
on this, both when coding, and when answering unnecessary questions for
others.

Jon

On 1/20/2022 11:13 AM, Jordan Brown wrote:

Places where from some angles you can see through the model are a dead
giveaway for convexity problems.  Setting convexity=2 in the
polyhedron() call fixes this.

I still think that we should set the default convexity to 10, to make
the vast majority of these problems go away with only a very small
performance hit.

I vote for a default convexity of 100.  We all waste far too much time on this, both when coding, and when answering unnecessary questions for others. Jon On 1/20/2022 11:13 AM, Jordan Brown wrote: > Places where from some angles you can see through the model are a dead > giveaway for convexity problems.  Setting convexity=2 in the > polyhedron() call fixes this. > > I still think that we should set the default convexity to 10, to make > the vast majority of these problems go away with only a very small > performance hit.
MM
Michael Marx
Fri, Jan 21, 2022 12:44 AM

High convexity has a preview performance hit on complex models.

-----Original Message-----
From: jon [mailto:jon@jonbondy.com]
Sent: Fri, 21 Jan 2022 03:18
To: OpenSCAD general discussion; Jordan Brown; nop head
Subject: [OpenSCAD] Re: debugging a polyhedron....

I vote for a default convexity of 100.  We all waste far too much time
on this, both when coding, and when answering unnecessary questions for
others.

Jon

On 1/20/2022 11:13 AM, Jordan Brown wrote:

Places where from some angles you can see through the model are a dead
giveaway for convexity problems.  Setting convexity=2 in the
polyhedron() call fixes this.

I still think that we should set the default convexity to 10, to make
the vast majority of these problems go away with only a very small
performance hit.


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

--
This email has been checked for viruses by AVG.
https://www.avg.com

High convexity has a preview performance hit on complex models. > -----Original Message----- > From: jon [mailto:jon@jonbondy.com] > Sent: Fri, 21 Jan 2022 03:18 > To: OpenSCAD general discussion; Jordan Brown; nop head > Subject: [OpenSCAD] Re: debugging a polyhedron.... > > I vote for a default convexity of 100. We all waste far too much time > on this, both when coding, and when answering unnecessary questions for > others. > > Jon > > On 1/20/2022 11:13 AM, Jordan Brown wrote: > > Places where from some angles you can see through the model are a dead > > giveaway for convexity problems. Setting convexity=2 in the > > polyhedron() call fixes this. > > > > I still think that we should set the default convexity to 10, to make > > the vast majority of these problems go away with only a very small > > performance hit. > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org -- This email has been checked for viruses by AVG. https://www.avg.com
J
jon
Fri, Jan 21, 2022 12:48 AM

How about a user-configurable preference.  If you want the default to be
2, that's great; if I want the default to be 30, that is also great.

:)

On 1/20/2022 7:44 PM, Michael Marx wrote:

High convexity has a preview performance hit on complex models.

-----Original Message-----
From: jon [mailto:jon@jonbondy.com]
Sent: Fri, 21 Jan 2022 03:18
To: OpenSCAD general discussion; Jordan Brown; nop head
Subject: [OpenSCAD] Re: debugging a polyhedron....

I vote for a default convexity of 100.  We all waste far too much time
on this, both when coding, and when answering unnecessary questions for
others.

Jon

On 1/20/2022 11:13 AM, Jordan Brown wrote:

Places where from some angles you can see through the model are a dead
giveaway for convexity problems.  Setting convexity=2 in the
polyhedron() call fixes this.

I still think that we should set the default convexity to 10, to make
the vast majority of these problems go away with only a very small
performance hit.


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

How about a user-configurable preference.  If you want the default to be 2, that's great; if I want the default to be 30, that is also great. :) On 1/20/2022 7:44 PM, Michael Marx wrote: > High convexity has a preview performance hit on complex models. > >> -----Original Message----- >> From: jon [mailto:jon@jonbondy.com] >> Sent: Fri, 21 Jan 2022 03:18 >> To: OpenSCAD general discussion; Jordan Brown; nop head >> Subject: [OpenSCAD] Re: debugging a polyhedron.... >> >> I vote for a default convexity of 100. We all waste far too much time >> on this, both when coding, and when answering unnecessary questions for >> others. >> >> Jon >> >> On 1/20/2022 11:13 AM, Jordan Brown wrote: >>> Places where from some angles you can see through the model are a dead >>> giveaway for convexity problems. Setting convexity=2 in the >>> polyhedron() call fixes this. >>> >>> I still think that we should set the default convexity to 10, to make >>> the vast majority of these problems go away with only a very small >>> performance hit. >> _______________________________________________ >> OpenSCAD mailing list >> To unsubscribe send an email to discuss-leave@lists.openscad.org >
FS
FF Systems
Fri, Jan 21, 2022 1:07 AM

A .ini file (xml text) could hold all of the system "defaults" like
convexity, $fn, etc... .  Place it in the same folder as the openscad.exe
application.  Don't put anything in the ini file that could even remotely
result in a security conundrum and it shouldn't be all that onerous.  If
".ini" is some kind of taboo, then just make it an ".scad" file with a
"defaults()" function that allows any of the deemed defaults to be
configured.  Calling "defaults()" from the top of an scad model would
configure the defaults for the session.

On Thu, Jan 20, 2022 at 6:48 PM jon jon@jonbondy.com wrote:

How about a user-configurable preference.  If you want the default to be
2, that's great; if I want the default to be 30, that is also great.

:)

On 1/20/2022 7:44 PM, Michael Marx wrote:

High convexity has a preview performance hit on complex models.

-----Original Message-----
From: jon [mailto:jon@jonbondy.com]
Sent: Fri, 21 Jan 2022 03:18
To: OpenSCAD general discussion; Jordan Brown; nop head
Subject: [OpenSCAD] Re: debugging a polyhedron....

I vote for a default convexity of 100.  We all waste far too much time
on this, both when coding, and when answering unnecessary questions for
others.

Jon

On 1/20/2022 11:13 AM, Jordan Brown wrote:

Places where from some angles you can see through the model are a dead
giveaway for convexity problems.  Setting convexity=2 in the
polyhedron() call fixes this.

I still think that we should set the default convexity to 10, to make
the vast majority of these problems go away with only a very small
performance hit.


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


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

A .ini file (xml text) could hold all of the system "defaults" like convexity, $fn, etc... . Place it in the same folder as the openscad.exe application. Don't put anything in the ini file that could even remotely result in a security conundrum and it shouldn't be all that onerous. If ".ini" is some kind of taboo, then just make it an ".scad" file with a "defaults()" function that allows any of the deemed defaults to be configured. Calling "defaults()" from the top of an scad model would configure the defaults for the session. On Thu, Jan 20, 2022 at 6:48 PM jon <jon@jonbondy.com> wrote: > How about a user-configurable preference. If you want the default to be > 2, that's great; if I want the default to be 30, that is also great. > > :) > > > On 1/20/2022 7:44 PM, Michael Marx wrote: > > High convexity has a preview performance hit on complex models. > > > >> -----Original Message----- > >> From: jon [mailto:jon@jonbondy.com] > >> Sent: Fri, 21 Jan 2022 03:18 > >> To: OpenSCAD general discussion; Jordan Brown; nop head > >> Subject: [OpenSCAD] Re: debugging a polyhedron.... > >> > >> I vote for a default convexity of 100. We all waste far too much time > >> on this, both when coding, and when answering unnecessary questions for > >> others. > >> > >> Jon > >> > >> On 1/20/2022 11:13 AM, Jordan Brown wrote: > >>> Places where from some angles you can see through the model are a dead > >>> giveaway for convexity problems. Setting convexity=2 in the > >>> polyhedron() call fixes this. > >>> > >>> I still think that we should set the default convexity to 10, to make > >>> the vast majority of these problems go away with only a very small > >>> performance hit. > >> _______________________________________________ > >> OpenSCAD mailing list > >> To unsubscribe send an email to discuss-leave@lists.openscad.org > > > _______________________________________________ > OpenSCAD mailing list > To unsubscribe send an email to discuss-leave@lists.openscad.org >
JB
Jordan Brown
Fri, Jan 21, 2022 3:08 AM

On 1/20/2022 5:07 PM, FF Systems wrote:

A .ini file (xml text) could hold all of the system "defaults" like
convexity, $fn, etc... .  Place it in the same folder as the
openscad.exe application.  Don't put anything in the ini file that
could even remotely result in a security conundrum and it shouldn't be
all that onerous.  If ".ini" is some kind of taboo, then just make it
an ".scad" file with a "defaults()" function that allows any of the
deemed defaults to be configured.  Calling "defaults()" from the top
of an scad model would configure the defaults for the session.

We already have a per-user preferences method that covers much of the need.

For per-model... we can perhaps have $ variables that control some
things, and then the users are free to set those however they desire -
perhaps by "include <defaults.scad>".  Note that a defaults() module can
never do the right thing because of variable scoping; any changes that
it made would go out of scope when it returns.

I've done a little with changing the default for convexity.  I think
I'll look at making it both a preference and a $ variable, with the $
variable getting its default from the preference.

On 1/20/2022 5:07 PM, FF Systems wrote: > A .ini file (xml text) could hold all of the system "defaults" like > convexity, $fn, etc... .  Place it in the same folder as the > openscad.exe application.  Don't put anything in the ini file that > could even remotely result in a security conundrum and it shouldn't be > all that onerous.  If ".ini" is some kind of taboo, then just make it > an ".scad" file with a "defaults()" function that allows any of the > deemed defaults to be configured.  Calling "defaults()" from the top > of an scad model would configure the defaults for the session. We already have a per-user preferences method that covers much of the need. For per-model... we can perhaps have $ variables that control some things, and then the users are free to set those however they desire - perhaps by "include <defaults.scad>".  Note that a defaults() module can never do the right thing because of variable scoping; any changes that it made would go out of scope when it returns. I've done a little with changing the default for convexity.  I think I'll look at making it *both* a preference and a $ variable, with the $ variable getting its default from the preference.