On 27.03.20 22:11, Jordan Brown wrote: > On 3/27/2020 10:21 AM, Torsten Paul wrote: >> echo(data.glyph[0].advance.x); > > Did OpenSCAD get general-purpose "named member" features > while I wasn't looking, or is this part of the > object-literals work? The core changes to make that work are mostly there, but for making it an actually usable feature, that is basically the object-literals (or whatever this changes into once the discussions are done). There could be a step between, were only built-ins would be able to return such a data structure. This needs some more eyes on it, ideally including people who have some experience with language design as with all those kind of changes, it's hard to modify once out in the wild. ciao, Torsten.

On 3/27/2020 2:19 PM, Torsten Paul wrote: > On 27.03.20 22:11, Jordan Brown wrote: >> On 3/27/2020 10:21 AM, Torsten Paul wrote: >>> echo(data.glyph[0].advance.x); >> Did OpenSCAD get general-purpose "named member" features >> while I wasn't looking, or is this part of the >> object-literals work? > The core changes to make that work are mostly there, > but for making it an actually usable feature, that > is basically the object-literals (or whatever this > changes into once the discussions are done). > > There could be a step between, were only built-ins > would be able to return such a data structure. Ah, thanks.  I skimmed the pull request too fast and misunderstood the scope.  (I looked at rcolyer's comment of October 16th 2019 where he says things like "o5 = cube(5)" and thought the scope was "geometric objects as values".) > This needs some more eyes on it, ideally including > people who have some experience with language > design as with all those kind of changes, it's > hard to modify once out in the wild. I'll see if I can take a look.  I'm not a real language design expert, but I can fake some of it. Is there anything other than the PR that I should be looking at?