Hi
Does the Standards Board have an opinion on the discussion about non-binary fractional circular primitives happening
here.
Personally, I don't have a problem with 3-16ringNN primitives.
(2021-04-28, 5:40)Chris Dee Wrote: [ -> ]Hi
Does the Standards Board have an opinion on the discussion about non-binary fractional circular primitives happening here.
Personally, I don't have a problem with 3-16ringNN primitives.
First of all, I'm not completely sure what fwcain is arguing against. In one place it was to use 1-16 plus 1-8 to make 3-16, and in another it was to use ring1 + ring2 + ring8 to make ring11. Based on a later post, I think that he was indicating that the latter was a mistake, and his argument is that only 1-16, 1-8, 1-4, 1-2, and 4-4 circular primitives should be allowed.
I personally strongly disagree with his argument, for a number of reasons:
- Having a bunch of ring primitives does not (in my opinion) hurt anything. They are all very simple files, and they have very precise constraints on what they should contain.
- Forcing part authors to combine 1-16 and 1-8 ring primitives to make a 3-16 primitive just makes all their lives miserable.
- Rounding means that combining 1-16 and 1-8 ring primitives to approximate a 3-16 primitive will always be an approximation, due to the necessary 45 degree rotation introducing √2 / 2 into the matrix. (I guess you could put them in the other order and have a 22.5 degree rotation, but that would also require rounding.)
- The official primitives reference never says to limit circular primitives to powers of two, and even explicitly includes 3-4disc and 3-4edge as examples. Additionally, it lists currently available primitives, so the large number of existing non-power-of-two primitives show up there without any mention that they should be avoided, which IMO is an implicit stamp of approval.
Due to 4 above, I would normally say that nothing needs to be changed. But the fact that this discussion has happened means that that is probably not a good option. I'd personally be happy to explicitly state that combining circular primitives with non 90 degree angles (for example 1-16 and 1-8) is
forbidden in official parts (due to point 3 above).
(2021-04-28, 17:24)Travis Cobbs Wrote: [ -> ] I'd personally be happy to explicitly state that combining circular primitives with non 90 degree angles (for example 1-16 and 1-8) is forbidden in official parts (due to point 3 above).
So would I.
I guess we could add a 'Primitives Regulations' section to the
LDraw.org Official Parts Library Specifications, but there is quite a lot to document.
(2021-05-01, 20:27)Chris Dee Wrote: [ -> ]So would I.
I guess we could add a 'Primitives Regulations' section to the LDraw.org Official Parts Library Specifications, but there is quite a lot to document.
Or we could just add the following text (or similar) to the Primitives Reference:
To avoid rounding errors, it is preferable to use existing (or create new) fractional circular primitives rather than rotate an existing primitive by anything other than 90 degrees. For example 3-16XXXX.dat is preferred to combining 1-8XXXX.dat with 1-16XXXX.dat rotated by 22.5 degrees.