mirror of
https://github.com/qelectrotech/qelectrotech-source-mirror.git
synced 2026-10-14 17:44:13 +02:00
ispyisail
9078e4e321
The scripting API (scripts and the assistant's live mode) reached a folio through diagrams(), which builds every folio, for every call. A folio's own data -- count, title, uuid, title block, border, wire defaults -- now comes from folios(), built or not; a call reading or changing a folio's items builds that folio, as showing it does (built()); selecting a symbol by uuid builds the folio the project database places it on. Search and replace and the continuity check still build every folio. DiagramContent worked out one wire of each potential when made, which follows potentials through folio reports and built the folios at the other end, for every caller; it is now worked out when first asked (Search and replace asks; a merge asks both sides, as before). tst_lazyfolios: on a .qetz, a script reads every folio's title, uuid and border building none (fails with diagrams()), and the symbols and wires of the folio with the most links to other folios as when built, building that folio only (27 before potentials were lazy). ctest 96/96 but modal_quit_regression in both modes: a race in that test (gdb interrupts the main thread inside malloc while the folio view repaints during the dialog, master too), 2 of 23 runs here, 0 of 10 on master, passing alone. 342 corpus projects save the same in all four arms, and export the same wiring list and wire numbers lazily. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>