With npm, it’s possible to do ‘npm install --prefix <someprefix>’ to install into any location
For global installs, setting the prefix to OpenSCAD’s global library folder would work automatically.
For local installs, we could combine this with setting the OPENSCADPATH environment variable.

For more into, see https://en.wikibooks.org/wiki/OpenSCAD_User_Manual/Libraries

-Marius

1i7 wrote

That would be great!
You can register yourself by using npms adduser command against the
openscad-modules registry. More details can be found in the github
repository.

Let me know if there are any problems or suggestions.

--
Sent from: http://forum.openscad.org/

1i7 wrote

Well, one point we should consider when thinking about this optimization is
the problem of name clashing and versioning conflicts. If you install all
modules in a common location the user would need to handle conflicting
module versions and also naming issues. In my opinion, the advantage of
having a modular systems, where I (as a module author and consumer) only
have to take care about my own dependencies (and not the one somebody else
is using in combination with my module) outperforms the advantage of saving
space. So, using "include <node_modules/fractals/fractals.scad>" inside a
module, as its described in the simple example, does this trick. Maybe it's
not a real feature of OpenSCAD, but it works and gives us the isolation for
naming and versioning issues :)

Beside that, I agree, that the look and feel is not as good as it could be.
The 'node_modules' part is just in the path because I didn't want to add
another command line parameter to the npm call. Maybe there is a better
solution that has a better presentation/usability but preserves the above
mentioned property?

--
Sent from: http://forum.openscad.org/

kintel wrote

Hi Marius, thx for your feedback. I just answered 1i7 regarding the
suggestion to install modules globally... Right now, I think there is one
big advantage in installing the modules in the NPMs preferred hierarchy
under a local folder, whićh is, it helps us to avoid naming and versioning
issues. A more detailed explanation can be found in the reply to 1i7
question.

In my opinion having this ability can speed up the development of modules,
as module authors itself doesn't need to be concerned about the context of
their module usage, they just need to be concerned about their module and
their dependencies.

--
Sent from: http://forum.openscad.org/

I believe the existing of a standard package manager will prompt owners to
write libs with specific names.

Gadgetmind wrote

$  Runsun Pan, PhD $ libs: scadx , doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), editor of choice: CudaText  ( OpenSCAD lexer ); $ Tips ; $ Snippets

--
Sent from: http://forum.openscad.org/

I worried about this quite a bit while working on scad_bundler (a
different package manager).

This is actually less of an issue than it appears if libraries are built
well.  See the Conventions listed on the github page for scad_bundler,
the 3rd bullet and later is meant to address this and it seems to work
pretty well (so far) in practice.

https://github.com/lostapathy/scad_bundler#draft-conventions

On 03/12/2018 06:59 PM, runsun wrote:

Interesting. I didn't know OpenSCAD has so-called 'public' var and 'private'
var. I really can't figure this out: Suppose I have 3 scads file:

============================
private.scad
abc="private";

public.scad
include <private.scad>

function f()= abc; // <=== use the private var
abc = "public"; // <== define the public var

user.scad
include <public.scad>

echo( private= f());
echo( public= abc);

Following your description on the "#draft-conventions", this will give

ECHO: private = "private"
ECHO: public = "public"

But it's not. Can you show an example ?

lostapathy wrote

$  Runsun Pan, PhD $ libs: scadx , doctest , faces ( git ), offline doc ( git ), runscad.py ( 2 , git ), editor of choice: CudaText  ( OpenSCAD lexer ); $ Tips ; $ Snippets

--
Sent from: http://forum.openscad.org/

In your case - private.scad gets (effectively) inlined into public.scad

• because it uses include instead of use.

So try this example:

###############################
~/scad_example $ cat private.scad
module priv_mod() {
  echo("I'm private");
}

###############################
~/scad_example $ cat public.scad
use <private.scad>

module pub_mod() {
  priv_mod();
  //also do other things
}

###############################
~/scad_example $ cat user.scad
use <public.scad>

echo("calling pub_mod()");
pub_mod();
echo("calling priv_mod()");
priv_mod();

###############################
~/scad_example $ openscad-nightly user.scad  -o blank.png

ECHO: "calling pub_mod()"
ECHO: "I'm private"
ECHO: "calling priv_mod()"
WARNING: Ignoring unknown module 'priv_mod'.
Compiling design (CSG Products normalization)...

###############################

I this case - user.scad is use'ing public.scad.  So while public.scad is
able to call priv_mod() on it's behalf, user.scad can't call that module
directly.

Joe

On 03/12/2018 08:00 PM, runsun wrote:

I was trying to illustrate that user.scad doesn't know about priv_mod()
at all - it's scope ends in prublic.scad, if you will. But what that
means is we can have more than one priv_mod() in different files, so
long as the same file doesn't use/include them both directly.

This is getting hard for email, so I put it on github
https://github.com/lostapathy/openscad-var-scope

I added a public2.scad and private2.scad.  Both private.scad and
private2.scad have a module named priv_mod(), but each can only be
called from the file that use'd them.  And neither leaks into
user.scad's namespace.

$ openscad user.scad  -o blank.png
ECHO: "calling pub_mod()"
ECHO: "I'm private"
ECHO: "calling pub2_mod()"
ECHO: "I'm private 2"
ECHO: "calling priv_mod()"
WARNING: Ignoring unknown module 'priv_mod'.

Hopefully that illustrates it better.  I haven't pushed this super hard
in any of my code yet, but it seems to work.

Joe

On 03/12/2018 09:17 PM, runsun wrote: