Skip to content

fix(print): randafsnijding bij posters alleen na expliciete keuze (#855) - #869

Closed
oscar981-op wants to merge 1 commit into
OpenAEC-Foundation:mainfrom
oscar981-op:fix/tauri-poster-crop-855
Closed

oscar981-op wants to merge 1 commit into
OpenAEC-Foundation:mainfrom
oscar981-op:fix/tauri-poster-crop-855

Conversation

@oscar981-op

@oscar981-op oscar981-op commented Oct 30, 2026 •

Copy link
Copy Markdown
Contributor

De posterindeling sneed kleine afrondingsverschillen automatisch af om minder vellen te gebruiken. Deze opvolging van #855 maakt dat een expliciete keuze: standaard blijft alle inhoud behouden, ook als daardoor extra vellen nodig zijn.

Het printvenster heeft nu Kleine buitenranden afsnijden, met de limiet zichtbaar erbij. De optie geldt alleen voor de huidige afdruk en staat bij heropenen weer uit. De maximale afsnijding is 0,75 mm per buitenrand, onafhankelijk van het aantal tegels. De schaal, overlap, printermarges en strook voor snijtekens blijven gerespecteerd. Een exact A1 op randloze, staande A3 bij 100%, zonder overlap/tekens/labels, gebruikt standaard zes vellen; met de optie vier, met 0,5 mm afsnijding aan de twee lange buitenranden. De teksten zijn vertaald in alle 40 talen.

Validatie:

  • De regressie faalt op de oorspronkelijke code: standaard twee tegelrijen in plaats van drie.
  • 25 gerichte postercontroles slagen, inclusief volledige paginadekking over schalen/standen en de vaste afsnijlimiet; echte PDF-inhoud wordt vergeleken met het voorbeeld voor beide keuzes.
  • Volledige suite: 6.185 geslaagd, 19 overgeslagen, nul fouten (6.204 totaal).
  • TypeScript, Vite-productiebuild en diffcontrole slagen.
  • Chromium/PDF.js: standaard uit, 6/4 vellen, terugzetten van de navigatie, gerenderde tegelpixels, echte PDF-downloads met 6/4 A3-pagina's en opnieuw openen met de optie uit. Schermbeeld gecontroleerd. De eerste browserstart had een timeout tijdens gelijktijdige builds; de herhaling slaagt.

Versie 2026.44.19 volgt het repositorybeleid. Concept totdat de controles en installerworkflow zijn beoordeeld. Deze branch erft de bekende macOS-watermerkregressie uit main; de afzonderlijke correctie staat in #868. De posterpreview-uitbreiding #866 blijft een aparte wijziging. Fysieke printer en native Windows/WebKit-bediening zijn niet lokaal getest. Releases blijven concept; geen publicatie.

Externe controles op d2e595b: macOS-CI faalt op de ongewijzigde watermerkdecoder uit main (de correctie #868 heeft inmiddels alle vijf upstreamcontroles doorstaan). Andere CI-builds/rendercontrole lopen nog.

Installerworkflow https://github.com/oscar981-op/open-pdf-studio/actions/runs/40977694608 is geëindigd als cancelled. Web en MCPB zijn geslaagd; Windows faalde wegens de ontbrekende Azure-signingendpoint en macOS bij certificaatimport. Android en Linux werden geannuleerd, dus hun installercontrole is onvolledig. Deze werker heeft geen annulering aangevraagd; de oorzaak is niet vastgesteld. Release v2026.44.19 is gecontroleerd als concept op de exacte commit. Vereiste review en upstream-mergebevoegdheid ontbreken nog.

@DutchSailor DutchSailor mentioned this pull request Oct 31, 2026
@DutchSailor

Copy link
Copy Markdown
Collaborator

Thank you @oscar981-op! This is released in 2026.44.5 as one squashed commit with you as the author (1d4945f, merged through #915).

Review fixes:

  • Keep the per-tile rounding allowance from main as the default instead of
    replacing it with an opt-in fixed 0.75 mm bound; the opt-in default
    brought back the 3 x 3 (A1 on A3) and 5 x 5 (A0 on A4) of [Bug]: Unexpected tile count calculation and scaling in Poster printing mode #855.
  • rasterLangs keeps its signature and default; randAfsnijden false is the
    strict full-coverage grid. The existing [Bug]: Unexpected tile count calculation and scaling in Poster printing mode #855 tests stay unchanged.
  • The hint shows the actual largest crop of the print (posterRandAfsnijding,
    computed as if the option were on) instead of a fixed maximum, which was
    wrong for grids of more than two tiles.
  • Format the hint number with the locale decimal separator (toonGetal).
  • Tests for the default, the strict mode, full edge coverage over scales,
    orientations, overlap and cut marks, the per-axis crop bound and the
    reported crop; print = preview check for the strict mode.
  • Dropped the browser end-to-end script: it assumed the opt-in default.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants