Well, Happy Birthday Edward! ;-) Regards, Jon N
On Sunday, August 13, 2023 at 10:31:16 AM UTC+1 Edward K. Ream wrote: > Today is my 74th birthday. This Engineering Notebook post is my birthday > present to myself. > > > For the last week, I have been trying to make paste-retaining-clones work. > Unit tests show this command improperly computes v.parents arrays in the > pasted outline. > > > This morning I saw the way forward: > > > *Aha!* paste-retaining-clones must recompute *all* v.parents arrays in > the entire outline! > > > I won't attempt a formal proof, but I am sure this Aha is correct. Note > that the paste-node command does *not *require the complications > described next. > > > *Recomputing v.parents* > > > But how to recompute v.parents? > > > - We can't use Position generators because they all use v.parents! > - We can't use VNode generators because they don't yield VNodes in outline > order. > > > *Aha! *Define a *c.all_vnode_positions* generator. This *hybrid generator* > will > yield VNodes in the same order as p.all_positions *using neither > v.parents nor Positions.* > > > Is it possible to write this generator? Yes, it is. The generator will > maintain an explicit stack just like Position.stack. *Note*: > c.all_vnode_positions must yield the hidden root VNode. > > > *Recomputing all v.parents* > > > Supposing the new generator exists, recomputing all v.parents arrays is > straightforward. Something like this (untested): > > > for v in c.all_vnode_positions(): > > v.parents = [] > > for v in c.all_vnode_positions(): > > for child in v.children: > > child.parents.append(v) > > > Yes, a one-pass solution is possible. It *might* be faster. > > > *Summary* > > > *Aha!* The paste-retaining-clones command must recompute *all *v.parents > ivars. This recomputation is sound. The link correction in > c.checkVnodeLinks is not. > > > Leo can use neither Position nor VNode generators to do the recomputation. > > > *Aha!* A new generator, *c.all_vnode_positions*, will yield all VNodes in > outline order using neither v.parents nor Positions. > > > Given this generator, updating all v.parents ivars will be straightforward. > > > Edward > > > P.S. Let us define a *hybrid generator *as a generator yielding VNodes in > the same order as some Position generator *without *using Positions. > > > Could Leo replace Position generators with hybrids? I think not. For > example, please take a look qtree.full_redraw. It would not be fun to > eliminate Positions there. > > > In any event, Leo's Position class is never going away :-) > > > EKR > -- You received this message because you are subscribed to the Google Groups "leo-editor" group. To unsubscribe from this group and stop receiving emails from it, send an email to leo-editor+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/leo-editor/667252b3-0580-4a72-ab5e-29154440d1d8n%40googlegroups.com.
