P12 native trials

Trial 07 establishes P12 native acceptance on the recorded macOS build. The lead review confirms numerical, visual, reopened-project, and owned-process checks. The probe records automated completion separately from this later review.

Native rendering build

Trial 07 uses native source 9c920334e338dab4908aa4dabdfae22803e70411 and its matching generated RIPS client. The complete native patch preserves the changes from 551dc02e19a1eb75ae462a0313f7a9a3101c2f45. These repairs are required for the reviewed summary image export workflow. The upstream 2026.9 version alone does not identify these changes.

The first build command and first build log preserve the initial renderer repair. The second build command and second build log preserve the title visibility and label painter repair. Both earlier build records report exit code zero. The final build command and final build log record the completed title repair. The repair 02 installation records the earlier application wheel. The repair 03 installation records application source b4c414beee91a0fa419f04e7c7874d2b2108a5a7 with the selected one-cell plot layout. This record describes local acceptance provenance and does not provide a released native distribution.

Trial 01

Trial 01 failed during the first native result load. It did not reach native numerical readback, image capture, or project reopening. This trial does not establish P12 native acceptance.

The command uses the reviewed installed interpreter and two generated P11 results. The installation record identifies application source 38e5aa8d9eca0c81e72ffdbdddfcf561e8735cbe and its matching generated RIPS client. The environment records native source 551dc02e19a1eb75ae462a0313f7a9a3101c2f45 and the probe checkout. The probe copy preserves the exact trial code. The probe checkout differs from the installed application source.

The production service accepted both immutable output bundles before native launch. The baseline bundle and scenario bundle preserve their complete result identities. The baseline dataset and scenario dataset preserve their accepted source values and geometry.

The first ResultsService.load returned execution_failed with mutation effect unknown. Its message was ResInsight call failed: No result returned from Method. The client error preserves the captured command output. The failure record identifies the failed stage and missing image evidence. Intermediate native inventory was not captured, so this trial does not identify the failed call within native loading. The native log contains no more specific import failure.

Source inspection supports a duplicate summary import explanation. RicfLoadCase uses the normal reader settings from RiaPreferencesGrid, which enable companion summary import. RimSummaryCaseMainCollection::createSummaryCasesFromFileInfos skips an existing summary filename. RimProject_importSummaryCase::execute then returns no object when it creates no new case. This explanation does not replace intermediate runtime observations.

The connection records owned process 68680 with start marker 1788977632.649255. The public close receipt confirms owned termination after failure. The independent process check confirms that the process identifier was absent afterward. The native lane was released only after these checks.

Trial 02

Trial 02 used the rebuilt application wheel from source 917dd33a836aa0349d3620cf9e1d02cb8fb0a122. The command retained the same generated models and native application build. Both production result loads and the following rebind succeeded. The recorder then failed while serializing native observations, before writing its first native readback file. The client error reports RepeatedScalarContainer as unsupported JSON input. The exact probe identifies the completed service calls before that recorder failure. This attempt produced no images and did not complete P12 native acceptance.

The public close receipt confirms termination of the owned process after failure. The independent process check confirms that process 80450 was absent afterward. The recorder repair converts supported repeated scalar sequences to ordinary lists at their native read boundaries. It preserves all values and existing comparisons.

Trial 03

The third command used application source 917dd33a836aa0349d3620cf9e1d02cb8fb0a122 and corrected probe source a31748d9a385ffe88e3f4b1dc163eec11a1b48cf. The installed application, generated RIPS client, native build, and two generated results stayed unchanged from trial 02. The installation record, environment, and exact probe preserve that provenance. The command exited successfully and completed every programmed numerical and project check.

The numerical review records four complete native readbacks before and after project reopening. Each readback contains 300 active cells on a 10 × 10 × 3 grid. Each checks 7,200 corner coordinates, six property arrays containing 1,800 values, and three summary curves containing six report values. All source and native corner, property, and summary differences are exactly zero. The properties are PRESSURE, SWAT, and SGAS at two report times. The curves are FOPR, WBHP:PROD, and WBHP:INJ.

The native grid includes its initial report on January 1, 2015. The accepted reports are January 2 and January 3, at elapsed days 1 and 2.

The baseline native record and scenario native record retain every observed value and corner. The restored baseline record and restored scenario record repeat those checks after reopening. The baseline source files and scenario source files accompany their manifests. Each source folder contains the five unchanged files used by the trial. P11 supplies verified source units, since the pinned native API has no separate unit getter. These comparisons establish faithful native readback, not an independent simulator physics reference.

The pressure comparison ranges from 0 to 87.21044921875 psi for scenario minus baseline. The water saturation comparison ranges from -0.00006521493196487427 to 0. The producer pressure comparison ranges from 470.05029296875 to 498.1962890625 psi. Both images in each grid pair have identical native legends and cameras. Native water saturation legend endpoints retain the existing 1e-14 relative tolerance against requested endpoints.

The checkpoint contains both result identifiers. The probe saved the native project, removed its local copy, and restored the same path from the immutable project artifact. It reopened that project and rejected the old reference. The fresh bindings retain both exact results and their new application context. The saved project preserves the resulting native configuration.

The image review covers all five original PNGs at 1200 × 800 pixels. All four grid images show the grid, well labels, result identity, requested quantity, and final report date. Their pressure units remain in the observation metadata and are not visible in the native legends. The summary PNG has a white left strip and a mostly transparent region that displays as black. It has no readable curve, title, or axes, so complete native acceptance remains false. The applied plot receipt, observation, and provenance preserve this failed visual result.

The saved project shows child summary plot identifier -1 inside parent MultiPlot identifier 3. The native exporter treats -1 as every docked plot. The targeting repair resolves the supported plot.ancestor(rips.MultiPlot) and requires a nonnegative identifier before export. This targets one plot window without claiming that it fixes the rendering defect.

The public close receipt confirms owned termination. The independent process check confirms that process 84243 was absent afterward. The native lane was released before evidence review. The review script and review log preserve the checks applied to saved records.

Trial 04

Trial 04 used application source 527d402, native source 70399abb, and probe source b7c0ccc. The exact command records their full identifiers and the unchanged generated results. The native repair uses the existing plot page renderer for PNG output. The native patch, build command, and build log preserve that change. The installation record also records compiler, CMake, and matching generated client provenance.

The automated completion record reports success and leaves visual acceptance pending. The numerical review again finds zero source/native corner, property, and summary differences before and after reopening. All four grid images retain matching legends, cameras, result identifiers, quantities, and final report dates. The saved project contains both requested plots and both results. The review script and log retain the saved-record checks.

The well request creates scenario WBHP:PROD at 1200 × 800 pixels. The field request creates scenario FOPR at 1000 × 700 pixels using a fresh project context. The first inventory contains the well plot under parent window 3. The second inventory preserves that plot and adds the field plot under parent window 4. The second export succeeds while the first plot remains present. Both well and field edit receipts retain their exact numerical curves.

The visual review inspected all six original PNGs. Both summary images now contain readable curves and axes with units. Neither image shows its configured result title, their date labels overlap, and each graph occupies only the upper-left canvas region. These remaining presentation defects prevent a complete visual acceptance claim. The title metadata review confirms that both child descriptions and all saved title visibility flags are present. The original images are preserved below.

The public close receipt confirms owned termination. The independent process check confirms that process 99165 was absent afterward. The native lane was released before further investigation.

Trial 05

Trial 05 used application source b4c414b, native source cd6450ab, and probe source b9e90a4. The exact command records their complete identifiers. The application sets the selected plot parent to one column and one row before export. The reviewed native patch preserves the complete plot page and correct date label scaling.

A recorder preflight attempt created the output directory before the probe required that directory to be new. The preflight error occurred before native launch. The recorder preserved its logs, removed that empty directory, and ran the unchanged command successfully. This preflight error does not indicate a product failure.

The completion record confirms automated success and leaves visual acceptance pending. The numerical review verifies all four native records with exactly zero geometry, property, and summary differences. The saved project reopens with both exact result identities. The first plot inventory and second plot inventory preserve parent 3 while adding parent 4. The review script preserves these saved-record checks.

The visual review inspects all six original PNGs. Both summary graphs now fill their single-cell pages with readable dates, units, and curves. The well image measures 1200 × 800 pixels, and the field image measures 1000 × 700 pixels. Both images still omit the configured result title, so complete visual acceptance remains pending. All four grid images retain their intended property, result, report, matching legend, and camera.

The public close receipt confirms owned termination. The independent process check confirms that process 10704 was absent afterward.

Trial 06 diagnostic

Trial 06 uses temporary native instrumentation to inspect missing title text. It does not change rendering behavior or establish final visual acceptance. The command requires native commit 630588f9dfc63e15ee7867134af5d18b6458f9a2 with the unchanged installed application. The environment records the expected and actual native commits before launch. The build command, build log, and diagnostic patch preserve its provenance.

The title diagnostics show white foreground text for both parent and child labels. Their text, page rectangles, and page-relative visibility are correct, and the painter has no clipping. The white text explains its absence against the white exported page. The completion record reports automated success with complete acceptance still false. The public close receipt confirms owned termination. The independent process check confirms that process 18351 was absent afterward.

Trial 07

Trial 07 uses the final reviewed native build at 9c920334e338dab4908aa4dabdfae22803e70411. The exact command uses the unchanged installed application and matching RIPS client. The complete native patch includes the title rendering repair and the separately reviewed cell containment correction. The temporary title diagnostics are removed. P12 checks result loading, numerical readback, plots, and restored projects. It does not establish P13 well export acceptance or independently validate simulator physics. The separate containment tests report 14 passing native checks. Their earlier regression run records five failures and one pass before repair.

The numerical review finds zero source/native differences in all four records before and after reopening. Each record retains 300 active cells, 7,200 corner coordinates, 1,800 property values, and six summary values. The saved project retains both result identities. The plot inventories preserve parent window 3 and add distinct parent window 4. The well request and field request retain their exact quantities and image dimensions.

The visual review records inspection of all six images. Both summary images show their result title, quantity, unit, dates, and full single-cell plot layout. All four grid images show the intended result, property, and report with matching legends and cameras. Grid pressure units remain in observation metadata and are not visible in native legends. Both the P12 agent and lead inspected all six images and accepted their content and layout.

The public close receipt confirms owned termination. The independent process check confirms that process 23148 was absent afterward.