i agree, that sounds better.
we just need somebody to actually implement it :)
On 22.09.23 23:34, Jordan Brown wrote:

On 23.09.23 00:06, pproj@posteo.de wrote:

That's 100% against the whole point of the experimental flag. It's
there so it's clear you can NOT rely on that feature for anything
next week.

ciao,
Torsten.

well,Thorsten.
 it is a good Point again.
Perhaps something more common like assert("experimental_off") could do?
On one hand, we do not want to restrict testing of new features. on the
other hand surprises for the authors of libraries mean avoidable work.
Many people including me use a dev version of everything if they can. so
it might be useful in these cases.  Perhaps such assertion could produce
a warning, not an error. just an Idea.

On 23.09.23 00:27, Torsten Paul wrote:

On 23.09.23 01:20, pproj@posteo.de wrote:

That seems a bit strong, but I see how some sort of warning about
this would be very useful. A bit similar to the "tainted" flag in
the Linux kernel but with a different focus.

Yep, exactly. So there's probably no perfect solution, simply
because the type of experimental features are very different,
e.g. the "sorted STL output" one is very specific but the lazy
union is really not finished and has various issues that may or
may not be solved eventually.

Giving the library a way of shouting out a specific warning
message might be one thing, maybe also adding a special menu button
for reporting issues.

So far, we've been asking for the "Library Info" content, which
has lots of technical info. I can imagine a reduced version could
help not only for reporting OpenSCAD issues, but also other types
of problems. This information does include a list of all available
experimental features along with there on/off state.

ciao,
Torsten.

On 9/22/2023 3:27 PM, Torsten Paul wrote:

Yes, sure.  Maybe it should be experimental(name) rather than
feature(name) to make that clearer.  But people will (and we want them
to!) build models using experimental features, and it seems like it
would be a good idea to be able to mark a model as depending on a
particular experimental feature.

Some experimental features have no semantic change on the model. For
example, manifold should not change it and the vbo changes only affect the
speed of rendering. So  assert("experimental_off") would break things for
no reason.

On Sat, 23 Sept 2023 at 02:43, Jordan Brown openscad@jordan.maileater.net
wrote:

On 9/23/2023 3:19 AM, nop head wrote:

Note that I was suggesting experimental(name); you would ask about the
experimental feature that you are interested in.

You could pointlessly assert that irrelevant experimental features are
on or off... don't do that.