deploy.

2026-09-13 19:45:24 +08:00
parent 397a7a8cc4
commit 2db889ad89
58 changed files with 3526 additions and 213 deletions
@@ -87,7 +87,68 @@
        <article data-uid="">
          <h1 id="release-note">Release Note</h1>

-<h2 id="hinc-packages-32">HiNc Packages 3.2</h2>
+<h2 id="hinc-packages-version-3224">HiNc Packages Version 3.2.24</h2>
+<p>Correctness and coverage that landed after 3.2.7. The 3.1.175 → 3.2 line jump is the
+3.2.7 entry below; the long form is still
+<a href="upgrading-to-3.2/index.html">Upgrading from 3.1.175 to 3.2</a>.</p>
+<div class="IMPORTANT">
+<h5>Important</h5>
+<p><strong>Adoption status.</strong> Verification is still in progress across most areas.</p>
+<p><strong>Do not put 3.2 NC optimization into production.</strong> It is not finished on this line.
+Work that depends on optimization should stay on the 3.1 line, serviced as
+<code>3.1.175.&lt;patch&gt;</code>. The Blazor front end tracks the 3.2 packages, so it is not a way
+to stay on the 3.1 behaviour.</p>
+<p><strong>Do not put SoftNc canned-cycle indexing with rotary words into production.</strong> SoftNc is
+the default engine, and its G81–G89 family does not take A / B / C words on a repeat
+block: a rotary-only repeat does not fire the cycle, a mixed block drops the rotary
+word, and under G68.2 the hole follows the table. Programs that index a rotary axis that
+way should run with
+<a class="xref" href="../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_EnableSoftNcRunner">EnableSoftNcRunner</a> set <code>false</code> — the
+HardNc fallback indexes those blocks the way a control does. See
+<a href="upgrading-to-3.2/brand-nc-language-coverage.html#fanuc-and-iso-common">Brand NC language coverage</a>.</p>
+</div>
+<ul>
+<li><strong>Whole-program throughput against the last 3.1 set is 2.5×–18.8×</strong> in steps per
+second (paired on one 32-logical-processor Linux workstation; the ratio grows with
+program length and mesh fineness). The six-core ceiling that used to cap the 3.1
+pipeline is about 6% of that gap. Figures and conditions:
+<a href="upgrading-to-3.2/performance-and-footprint.html">Performance and footprint</a></li>
+<li><strong><code>M02</code> / <code>M30</code> / <code>END PGM</code> is a reset edge on SoftNc.</strong> Tool-length compensation
+(including TCP), tilt, cutter-radius compensation and canned cycles cancel on the <em>next</em>
+block; the end-of-program block itself still runs under the modal state it executed with.
+HardNc still carries <code>G43.4</code> across <code>M02</code>. See
+<a href="upgrading-to-3.2/results-that-change-on-upgrade.html">Results that change on upgrade</a></li>
+<li><strong>An omitted H is no longer a silent zero</strong> — it takes the equipped tool's offset row
+first, then warns <code>Comp-ToolHeight--NoToolForOmittedH</code> / <code>--NoToolForOmittedHRtcp</code>. A
+vacant table row warns <code>Comp-ToolHeight--RowMissing</code></li>
+<li><strong>Cutter-radius compensation runs in the selected G17 / G18 / G19 plane.</strong> An
+interference path that used to swap sides silently now reports
+<code>RadiusComp--Interference</code></li>
+<li><strong>Siemens <code>ROTS</code> / <code>AROTS</code> are simulated</strong> into the tilt chain in Sinumerik RPY order.
+<code>SCALE</code> / <code>MIRROR</code> / <code>CROTS</code> remain recognized, not simulated</li>
+<li><strong>Heidenhain <code>LN</code> vector blocks</strong> resolve into rotary axes through
+<a class="xref" href="../api/Hi.NcParsers.LogicSyntaxs.Heidenhain.HeidenhainLnOrientationSyntax.html">HeidenhainLnOrientationSyntax</a></li>
+<li><strong>Fanuc <code>G54.1 Pn</code> (also written <code>G54 Pn</code>) and <code>G59.1</code>–<code>G59.9</code></strong> resolve additional work
+offsets. A selected row nobody has entered reports <code>Coord-WorkOffset--AdditionalZero</code></li>
+<li><strong>Runner feedrate members rename to <code>CommandedClFeedrate_mmds</code>.</strong> Physics follows the
+equipped tool's tip via
+<a class="xref" href="../api/Hi.MachiningSteps.MachineMotionStep.html#Hi_MachiningSteps_MachineMotionStep_ActualTipFeedrate_mmds">ActualTipFeedrate_mmds</a>. A project-level
+<a class="xref" href="../api/Hi.NcParsers.Dependencys.Generic.RadiusOffsetBasis.html">RadiusOffsetBasis</a> decides what a tool-house
+refresh writes into the D column — see
+<a href="../anatomy/general-setup/controller/program-data/tool-offsets.html#the-radius-basis">Tool Offsets</a></li>
+<li><strong><code>GetMillingEquipment</code> → <code>GetMachiningEquipment</code>.</strong> <code>CodeXyzabcMachineTool</code> is absorbed
+into <a class="xref" href="../api/Hi.NcMech.Xyzabc.GeneralXyzabcMachineTool.html">GeneralXyzabcMachineTool</a> (the old XML name still loads). The
+equipment splits into authored
+<a class="xref" href="../api/Hi.Machining.MachiningEquipmentUtils.SetupEquipment.html">SetupEquipment</a> and runtime
+<a class="xref" href="../api/Hi.Machining.MachiningEquipmentUtils.MachiningEquipment.html">MachiningEquipment</a></li>
+<li><strong>Close joins the <a class="xref" href="../api/Hi.MachiningProcs.ProjectFileBusyException.html">ProjectFileBusyException</a> gate.</strong>
+<a class="xref" href="../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_LoadProject_">LoadProject</a> and
+<a class="xref" href="../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_ReloadProject_">ReloadProject</a> take a message sink for
+load-time diagnostics</li>
+<li><strong><a class="xref" href="../api/Hi.NcOpt.NcOptOption.html#Hi_NcOpt_NcOptOption_OmitLeadingZero">OmitLeadingZero</a></strong> is an output style the project can pin
+(<code>.5</code> vs <code>0.5</code>)</li>
+</ul>
+<h2 id="hinc-packages-version-327">HiNc Packages Version 3.2.7</h2>
 <h3 id="at-a-glance">At a glance</h3>
 <p>3.2 is largely one piece of work: the NC interpreter that reads a controller program is now composed
 from configurable parts rather than written into one class, and most of this release is the brand
@@ -152,7 +213,7 @@ depends on optimization should stay on the 3.1 line, which is serviced as <code>
 Blazor front end tracks the 3.2 packages, so it is not a way to stay on the 3.1 behaviour.</p>
 </div>
 <h3 id="what-changed">What changed</h3>
-<p><strong>This one entry covers everything since 3.1.175.</strong> The 3.1 line is closed at the 3.1.175 package
+<p><strong>This 3.2.7 entry covers everything since 3.1.175.</strong> The 3.1 line is closed at the 3.1.175 package
 set and is serviced only as <code>3.1.175.&lt;patch&gt;</code>; <code>master</code> moved all ten packages onto the <strong>3.2</strong> line
 on 2026-08-24 and restarted their build counters. So a 3.2 build number starts low, the two
 counters are not comparable, and the gap between the last 3.1 number a feed served and the first
@@ -107,11 +107,13 @@ data, <code>DEF REAL/INT</code> declarations lower into assignments, and <code>$
 to <a class="xref" href="../../api/Hi.NcParsers.Dependencys.Siemens.SiemensFrameTable.html">SiemensFrameTable</a>. Any other <code>$</code>-variable is recorded
 with an unsupported note rather than dropped.</li>
 <li><strong>Five axis</strong> — <a class="xref" href="../../api/Hi.NcParsers.LogicSyntaxs.Siemens.SiemensProgrammableFrameSyntax.html">SiemensProgrammableFrameSyntax</a> simulates
-<code>TRANS</code> / <code>ATRANS</code> / <code>ROT</code> / <code>AROT</code> (with <code>RPL=</code>) into the tilt-transform chain in Sinumerik RPY
-order; <a class="xref" href="../../api/Hi.NcParsers.LogicSyntaxs.Siemens.SiemensTraoriSyntax.html">SiemensTraoriSyntax</a> makes <code>TRAORI</code> a real RTCP mode,
-the sibling of ISO G43.4, with <code>TRAFOOF</code> handing the offset back;
+<code>TRANS</code> / <code>ATRANS</code> / <code>ROT</code> / <code>AROT</code> (with <code>RPL=</code>) and the solid-angle forms <code>ROTS</code> / <code>AROTS</code> into the
+tilt-transform chain in Sinumerik RPY order (<code>ROTS</code> / <code>AROTS</code> behave as <code>ROT</code> / <code>AROT</code>: at most two
+angles, first-named axis first); <a class="xref" href="../../api/Hi.NcParsers.LogicSyntaxs.Siemens.SiemensTraoriSyntax.html">SiemensTraoriSyntax</a> makes
+<code>TRAORI</code> a real RTCP mode, the sibling of ISO G43.4, with <code>TRAFOOF</code> handing the offset back;
 <a class="xref" href="../../api/Hi.NcParsers.LogicSyntaxs.Siemens.SiemensCycle800TiltSyntax.html">SiemensCycle800TiltSyntax</a> decodes <code>CYCLE800</code>'s MODE bits for
-all four swivel modes. <code>ROTS</code> / <code>SCALE</code> / <code>MIRROR</code> are recognized and reported, not simulated.</li>
+all four swivel modes. <code>SCALE</code> / <code>MIRROR</code> / <code>CROTS</code> are recognized and reported
+(<code>SiemensFrame--Unsupported</code>), not simulated.</li>
 <li><strong>Calls</strong> — L-prefixed and named subprogram calls resolve against
 <a class="xref" href="../../api/Hi.NcParsers.Dependencys.Generic.SubProgramFolderConfig.html">SubProgramFolderConfig</a> (<code>{name}.SPF</code>, then <code>.MPF</code>, then the
 bare name) and inline with their <code>P</code> repetition count; <code>M17</code> / <code>RET</code> pop a frame; <code>REPEAT</code> re-runs a
@@ -155,9 +157,13 @@ and <code>DL</code>. Datum handling follows TNC semantics: <code>CYCL DEF 247</c
 <em>additive</em> shift on top of it, composing as separate transform-chain entries instead of replacing
 each other, resolved against <a class="xref" href="../../api/Hi.NcParsers.Dependencys.HeidenhainDatumTable.html">HeidenhainDatumTable</a>. Arcs (<code>CC</code> pole
 plus <code>C</code> statement, <code>DR-</code> = CW, closed arc = full circle), <code>RL</code> / <code>RR</code> / <code>R0</code> radius compensation,
-the <code>M126</code> / <code>M127</code> rotary-wrap state, <code>M140 MB</code> retract, and <code>CYCL DEF 32 TOLERANCE</code>. A <code>C</code> block
-never states its own centre: each in-plane component comes from the <code>CC</code> block's own axis word,
-else from the previous <code>CC</code> section's same axis, else from the arc's own start point. A bare <code>CC</code>
+the <code>M126</code> / <code>M127</code> rotary-wrap state, <code>M140 MB</code> retract, and <code>CYCL DEF 32 TOLERANCE</code>.
+<a class="xref" href="../../api/Hi.NcParsers.LogicSyntaxs.Heidenhain.HeidenhainLnOrientationSyntax.html">HeidenhainLnOrientationSyntax</a> resolves an <code>LN</code> vector
+block into rotary axes through the same RTCP machinery the rotary-word programs use: <code>T</code> present
+plus <code>M128</code> / <code>FUNCTION TCPM</code> keeps the tool on T; <code>T</code> absent plus RTCP holds the tool on the
+surface normal N; RTCP inactive ignores T (<code>Orientation-Vector--IgnoredNoTcpm</code>), as the control
+does. A <code>C</code> block never states its own centre: each in-plane component comes from the <code>CC</code> block's
+own axis word, else from the previous <code>CC</code> section's same axis, else from the arc's own start point. A bare <code>CC</code>
 is the one spelling that states all of them at once — it takes the last programmed position, read
 at the <code>CC</code> block, and replaces the modal centre rather than inheriting it. A centre that lands on
 the arc's own start point leaves the block with no radius, so it warns
@@ -204,7 +210,9 @@ fabricated value. <strong>FN 9–12 conditional jumps execute</strong>, with a (
 <code>STAY</code> / <code>MOVE</code> / <code>TURN</code> positioning; VECTOR structurally captured; EULER / POINTS / RELATIV /
 AXIAL / PROJECTED consumed and warned with the previous tilt retained, so a <code>PLANE AXIAL B+45</code> B
 word can never be mistaken for a rotary axis command), <code>FUNCTION TCPM</code>, and real <code>M128</code> / <code>M129</code>
-tool-centre-point control.</li>
+tool-centre-point control. A centre-referenced <code>FUNCTION TCPM REFPNT</code> (<code>TIP-CENTER</code> /
+<code>CENTER-CENTER</code>) is recorded and reported (<code>Orientation-RefPoint--CntNotSimulated</code>) rather than
+silently read as TIP-TIP.</li>
 <li><strong>Cycles and calls</strong> — <code>CYCL DEF 2xx</code> bodies with their Q parameters mirrored into the block
 assignments, cycles 200 / 232 / 251 / 252 / 253 mapped onto the shared G81 / G82 / G83 slots,
 <code>CYCL CALL</code> / <code>CYCL CALL POS</code> / <code>M99</code> / <code>M89</code> splitting call-once from modal firing, <code>CALL LBL</code>
@@ -241,6 +249,20 @@ codes before the mode syntaxes consume them.</li>
 <code>WHILE[..]DO m / END m</code> with a bounded-loop watchdog, and position and tool-offset system variables.
 <code>M98 P_ L_</code>, <code>M198</code> external call, <code>M99</code> return and <code>M99 P{seq}</code> early return; <code>G65</code> one-shot macro
 call with A–Z → <code>#1</code>–<code>#26</code> argument binding, and <code>G66</code> / <code>G67</code> modal macro.</li>
+<li><strong>Additional work coordinate systems</strong> — <code>G54.1 Pn</code>, also written <code>G54 Pn</code>, resolves against the
+extended table (P1–P48). A selected row nobody has entered reports <code>Coord-WorkOffset--AdditionalZero</code>.
+<code>G59.1</code>–<code>G59.9</code> have a provider on the Fanuc, Mazak and Syntec presets. See
+<a href="../../anatomy/general-setup/controller/program-data/work-coordinates.html">Work Coordinates</a> and
+<a href="../../technique/nc-dialects/controller-iso.html">ISO core</a>.</li>
+<li><strong>Fanuc / Syntec unit codes and G05.</strong> <code>G71</code> is the metric spelling of <code>G21</code>; <code>G70</code> is the inch
+spelling of <code>G20</code> and reports <code>Unit--InchNotSupported</code> rather than silently switching units.
+<code>G05.1 Q1</code> / <code>Q0</code> is AICC / Nano Smoothing (recognized). Bare <code>G05 P{n}</code> is a different feature
+(HPCC) and is consumed without changing programmed coordinates, reported under the <code>Hpcc--*</code> ids.</li>
+<li><strong>Canned-cycle repeats do not take rotary words on SoftNc.</strong> The G73–G89 family expands XYZ / R /
+Q / F / P / K strokes. An A / B / C word on a repeat is not taken: a rotary-only repeat does not
+fire the cycle, a mixed block drops the rotary word, and under G68.2 the hole follows the table.
+<em>Not finished on this line, see Adoption status on the <a href="../index.html">release-note page</a>.</em> The
+HardNc fallback indexes those blocks the way a control does.</li>
 </ul>
 <h2 id="cross-brand">Cross-brand</h2>
 <ul>
@@ -304,6 +326,12 @@ re-read the host file by that path — the existence check always failed, so loo
 without looping. <code>RunNc</code> now registers the raw lines on
 <a class="xref" href="../../api/Hi.NcParsers.Dependencys.SystemWired.NcLineSourceDependency.html">NcLineSourceDependency</a> and
 <a class="xref" href="../../api/Hi.NcParsers.EvaluationSyntaxs.LabelScanUtil.html">LabelScanUtil</a> reads memory first, disk second.</li>
+<li><strong><code>M02</code> / <code>M30</code> / <code>END PGM</code> is a reset edge.</strong> SoftNc models program-end as the edge <em>between</em> that
+block and the next, so a concatenated multi-program file does not keep TCP, tilt, CRC or a canned
+cycle alive into the next program. Each modal owner writes an explicit cancel section on the
+successor (G49 / G69 / G40 / G80); the end-of-program block itself still runs under the modal state
+it executed with. HardNc still carries G43.4 across <code>M02</code> — see
+<a href="results-that-change-on-upgrade.html">Results that change on upgrade</a>.</li>
 </ul>

 </article>
@@ -225,6 +225,14 @@ everything else falls back to brand G-code, so an exotic brand extension can nev
 <td><code>StateActRunner.Feedrate_mmds</code> / <code>Feedrate_mmdmin</code>, <code>ActFeedrate.Feedrate_mmds</code> / <code>Feedrate_mmdmin</code> (and <code>ActRapid</code>), <code>MachineMotionStep.Feedrate_mmds</code>, the <code>MachineMotionStep</code> constructor's <code>feedrate_mmds</code> parameter</td>
 <td><a class="xref" href="../../api/Hi.Numerical.Acts.StateActRunner.html#Hi_Numerical_Acts_StateActRunner_CommandedClFeedrate_mmds">CommandedClFeedrate_mmds</a> / <a class="xref" href="../../api/Hi.Numerical.Acts.StateActRunner.html#Hi_Numerical_Acts_StateActRunner_CommandedClFeedrate_mmdmin">CommandedClFeedrate_mmdmin</a>, <a class="xref" href="../../api/Hi.Numerical.Acts.ActFeedrate.html#Hi_Numerical_Acts_ActFeedrate_CommandedClFeedrate_mmds">CommandedClFeedrate_mmds</a> / <a class="xref" href="../../api/Hi.Numerical.Acts.ActFeedrate.html#Hi_Numerical_Acts_ActFeedrate_CommandedClFeedrate_mmdmin">CommandedClFeedrate_mmdmin</a>, <a class="xref" href="../../api/Hi.MachiningSteps.MachineMotionStep.html#Hi_MachiningSteps_MachineMotionStep_CommandedClFeedrate_mmds">CommandedClFeedrate_mmds</a>, <code>commandedClFeedrate_mmds</code></td>
 </tr>
+<tr>
+<td><code>GetMillingEquipment</code></td>
+<td><a class="xref" href="../../api/Hi.Machining.MachiningEquipmentUtils.IGetMachiningEquipment.html#Hi_Machining_MachiningEquipmentUtils_IGetMachiningEquipment_GetMachiningEquipment_">GetMachiningEquipment</a></td>
+</tr>
+<tr>
+<td><code>CodeXyzabcMachineTool</code> / <code>CodeXyzabcMachineToolUri</code></td>
+<td><a class="xref" href="../../api/Hi.NcMech.Xyzabc.GeneralXyzabcMachineTool.html">GeneralXyzabcMachineTool</a> (the old XML names still load)</td>
+</tr>
 </tbody>
 </table>
 <p>Four notes on that table.</p>
@@ -371,6 +379,13 @@ updated in lockstep.</li>
 It is a public <code>const</code>, so an assembly compiled against 3.1.175 already carries the old literal and
 keeps passing it — <code>Init</code> still accepts it — but a rebuild changes what it passes, and no font file
 is extracted to the working directory any more.</li>
+<li>The equipment splits into two faces. <a class="xref" href="../../api/Hi.Machining.MachiningEquipmentUtils.SetupEquipment.html">SetupEquipment</a> is
+the authored one — the only face a project file persists, reached as
+<a class="xref" href="../../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_SetupEquipment">SetupEquipment</a>.
+<a class="xref" href="../../api/Hi.Machining.MachiningEquipmentUtils.MachiningEquipment.html">MachiningEquipment</a> is the runtime face the runner,
+physics, collision and execution display read; it is rebuilt from the authored face at project
+assignment and at session boundaries, so a value written onto it is discarded rather than saved.
+See <a href="../../technique/api-foundations/getting-started.html">Getting started</a>.</li>
 </ul>
 <h2 id="8-defaults-and-gates-that-changed">8. Defaults and gates that changed</h2>
 <ul>
@@ -393,6 +408,13 @@ and scripts are skipped with <code>Script--NotLicensed</code>. An unlicensed ins
 <em>different simulation</em>, not an error. Calling the public API from your own application or session
 script needs no extra licence; composing the interpretation pipeline does. See
 <a class="xref" href="../../technique/nc-dialects/nc-parsing.html">NC Parsing Engine</a>.</li>
+<li><strong><a class="xref" href="../../api/Hi.MachiningProcs.MachiningActRunnerConfig.html#Hi_MachiningProcs_MachiningActRunnerConfig_RadiusOffsetBasis">RadiusOffsetBasis</a></strong> decides what a tool-house
+refresh writes into the D column:
+<a class="xref" href="../../api/Hi.NcParsers.Dependencys.Generic.RadiusOffsetBasis.html#Hi_NcParsers_Dependencys_Generic_RadiusOffsetBasis_CutterRadius">CutterRadius</a> (default — geometry plus wear,
+the only behaviour before the basis existed) or
+<a class="xref" href="../../api/Hi.NcParsers.Dependencys.Generic.RadiusOffsetBasis.html#Hi_NcParsers_Dependencys_Generic_RadiusOffsetBasis_ZeroBased">ZeroBased</a> (wear only, for a tool-centre
+CAM path). An older project refreshes exactly as before until the basis is changed. See
+<a href="../../anatomy/general-setup/controller/program-data/tool-offsets.html#the-radius-basis">Tool Offsets</a>.</li>
 <li>The four <code>SnapshotSyntax</code> entries in the Fanuc preset default to <code>IsEnabled = false</code>, so projects
 stop serializing enabled debug snapshots. A project saved by an earlier build keeps what it
 serialized until its pipeline list is refreshed from the current preset.</li>
@@ -87,7 +87,8 @@
        <article data-uid="Release-Upgrading-To-3.2">
          <h1 id="upgrading-from-31175-to-32">Upgrading from 3.1.175 to 3.2</h1>

-<p>These pages are the long form of the <a href="../index.html">3.2 release-note entry</a>. They exist because 3.2 is not
+<p>These pages are the long form of the 3.1.175 → 3.2 line — the 3.2.7 entry and the
+<a href="../index.html">3.2.24</a> incremental drop. They exist because 3.2 is not
 an increment on the last release most callers hold — it is the accumulation of everything that
 landed after the 3.1.175 package set, delivered in one step.</p>
 <p>Read them in order the first time. The two sections that decide whether an upgrade is a recompile or
@@ -98,8 +99,9 @@ capability you can adopt when you need it.</p>
 <h5>Important</h5>
 <p>Read the <strong>Adoption status</strong> note on the <a href="../index.html">release-note page</a> before planning around this.
 The short version: 3.2 is published for review, verification is still in progress across most
-areas, and <strong>NC optimization is not finished on this line</strong> — work that depends on it should stay
-on 3.1, serviced as <code>3.1.175.&lt;patch&gt;</code>.</p>
+areas, <strong>NC optimization is not finished on this line</strong>, and <strong>SoftNc canned-cycle indexing
+with rotary A/B/C words is not finished</strong> — work that depends on either should stay on 3.1,
+serviced as <code>3.1.175.&lt;patch&gt;</code>, or run with SoftNc off for the rotary-cycle case.</p>
 </div>
 <h2 id="pages">Pages</h2>
 <p>Ordered the way to read them: what the package numbers mean, then the two sections that decide whether the upgrade is a recompile or an afternoon, then the new capability, adopted when needed.</p>
@@ -142,6 +142,11 @@ fourth — an alarm on controllers strict about their least input increment. Dig
 scanned per word family over the played source texts with comment spans masked, floored at 3/3/0 and
 capped at 9, and threaded through the whole write path. The word-suppression tolerances and the F
 comparison grid derive from the resolved digits instead of the old fixed literals.</li>
+<li><strong>Leading zeros follow the source, and the project can pin the spelling.</strong>
+<a class="xref" href="../../api/Hi.NcOpt.NcOptOption.html#Hi_NcOpt_NcOptOption_OmitLeadingZero">OmitLeadingZero</a> is an output style: on, the writer drops a leading
+zero before the decimal (<code>.5</code> rather than <code>0.5</code>); off, it writes the zero. The option is carried on
+the project so two machines produce the same bytes, and a report names which style the source
+itself used.</li>
 <li><strong>Optimized NC is written back in the source file's encoding.</strong> NC play reads and optimized writes go
 through <a class="xref" href="../../api/Hi.Common.FileLines.FileUtil.html#Hi_Common_FileLines_FileUtil_DetectRoundTripEncoding_">DetectRoundTripEncoding</a> — BOM, then strict UTF-8, then
 Latin-1 — so an ANSI-family file (GBK, Big5, Shift-JIS) re-encodes to its original bytes instead of
@@ -134,6 +134,22 @@ they are the fastest way to find out what a run actually did.</p>
 <td>a G43.4 H word could not be resolved</td>
 </tr>
 <tr>
+<td><code>Comp-ToolHeight--NoToolForOmittedH</code> / <code>--NoToolForOmittedHRtcp</code></td>
+<td>a G43 / G43.4 with no H, no equipped tool number and no non-zero modal offset id</td>
+</tr>
+<tr>
+<td><code>Comp-ToolHeight--RowMissing</code></td>
+<td>the resolved offset id has no table row</td>
+</tr>
+<tr>
+<td><code>Comp-ToolHeight--RtcpOutlivesToolChange</code></td>
+<td>a tool change left <code>G43.4 Hm</code> alive for a different tool; the new tool keeps the previous tool's length</td>
+</tr>
+<tr>
+<td><code>StrokeLimit--Unconfigured</code></td>
+<td>stroke-limit checking is on but the chain's linear axes carry no travel limit; once per session at BeginSession</td>
+</tr>
+<tr>
 <td><code>RadiusComp--Interference</code></td>
 <td>a G41/G42 block whose compensated path runs against the programmed direction (an inner arc or groove smaller than the offset radius); a real control stops there with an interference alarm (Fanuc PS0041) while the simulation continues on the swapped-side path — on a tool-centre CAM path the radius offset row (D row) must hold the wear only, not the tool radius</td>
 </tr>
@@ -210,6 +226,22 @@ they are the fastest way to find out what a run actually did.</p>
 <td>the construct is recognized and consumed safely, but not simulated</td>
 </tr>
 <tr>
+<td><code>HeidenhainCyclCall--IncrementalNoReference</code> / <code>--IncrementalAfterMove</code></td>
+<td>a klartext <code>CYCL CALL POS</code> <code>I</code>-prefixed word has no previous call to measure from, or the tool moved between the two calls</td>
+</tr>
+<tr>
+<td><code>Orientation-RefPoint--CntNotSimulated</code></td>
+<td><code>FUNCTION TCPM REFPNT</code> with a centre-referenced tool point is not simulated — coordinates are read as TIP-TIP</td>
+</tr>
+<tr>
+<td><code>Orientation-Vector--IgnoredNoTcpm</code></td>
+<td>an <code>LN</code> tool vector arrived with RTCP inactive and was ignored, as the control ignores it</td>
+</tr>
+<tr>
+<td><code>Hpcc--NoOp</code> / <code>--HighSpeedCycleIgnored</code> / <code>--UnsupportedFunction</code> / <code>--UnevaluatedFunction</code> / <code>--MissingFunctionWord</code></td>
+<td>a bare <code>G05 P</code> HPCC selection; the programmed coordinates are unchanged, except <code>P10001</code>–<code>P10999</code> which is real machining the simulation cannot see</td>
+</tr>
+<tr>
 <td><code>DeclaredMCode--UnmodeledEffects</code></td>
 <td>a machine-declared OEM M-code occurred; no effects are simulated for it</td>
 </tr>
@@ -91,6 +91,23 @@
 correctness that is said too. Read the ratios rather than the absolute times: several campaigns were
 run on Debug builds or on small machines, deliberately, because a paired A/B on one machine answers
 &ldquo;did this get faster&rdquo; far more reliably than an unpaired Release number on a fast one.</p>
+<h2 id="against-the-last-31-set">Against the last 3.1 set</h2>
+<p>Paired on one 32-logical-processor Linux workstation, deployed stacks, <strong>whole</strong> NC programs, steps per
+second. The 3.1 side is HardNc (the 3.1 default); the 3.2 side is SoftNc (the 3.2 default). Throughput
+on this line is <strong>2.5×–18.8×</strong>; the ratio grows with program length and mesh fineness. Memory is mixed:
+a cutting-dominated long program peaked higher on 3.2, a large-surface short-NC job peaked lower.</p>
+<p>The six-core ceiling that used to cap the 3.1 pipeline is <strong>not</strong> where that gap comes from. On the
+same workstation, pinning 3.2's worker counts to the products of that ceiling moved a ~193,000-step
+play from 15.0 s to 15.9 s (+6%); at six available cores the version advantage is still 2.8×–3.8×,
+and at two cores 2.73×. 3.1 saturates from eight cores; 3.2 from twelve. On that fixture, 3.1 allocated
+6–8× more managed memory, and the 3.1 allocation climbed as the mesh got finer while 3.2 stayed
+almost flat.</p>
+<blockquote>
+<p><strong>Conditions.</strong> Deployed x64 stacks, one machine, whole programs, steps/s (not wall-clock — a
+defect that inflated 3.2 step counts on an earlier drop is fixed on this line). Core-count figures
+used <code>DOTNET_PROCESSOR_COUNT</code> plus CPU affinity, so they are &ldquo;restricted cores on a large machine&rdquo;,
+not a small one: memory bandwidth and cache stay those of the 32-LP host.</p>
+</blockquote>
 <h2 id="milling-physics-in-the-native-kernel">Milling physics in the native kernel</h2>
 <p>The per-step physics moved into <code>core.dll</code> in stages — engagement scan conversion, the force kernel,
 then the sequential temperature and wear chain. Measured as a same-day paired A/B, managed leg
@@ -166,11 +183,11 @@ parallel physics stage is about 3.3% of wall time</strong>, while the <strong>si
 about 77%</strong> — and that subtraction is single-worker as a correctness requirement, not as an
 oversight. Raising the force-worker count therefore buys nothing on any machine; the bottleneck
 moved rather than disappearing.</p>
-<p>What did change in the worker derivation is narrower than it sounds. An unmeasured six-core ceiling
-was removed, but it only ever governed the sweep workers, and only machines with <strong>nine or more
-logical processors</strong> see a different count; force workers are unchanged everywhere. The throughput
-benefit on such a machine has <strong>not</strong> been measured — the development machines are smaller — so this
-is a ceiling removal, not a claimed speedup.</p>
+<p>What did change in the worker derivation is narrower than it sounds. The six-core ceiling
+was removed, and on a 32-logical-processor machine that is about <strong>6%</strong> of the 3.1→3.2 wall-time gap
+measured above — force workers are unchanged everywhere, and the sweep-worker count only differs on
+machines with <strong>nine or more</strong> logical processors. The rest of the gap is allocation and the rest of
+the pipeline, not the extra cores.</p>
 <p>Queue depths became fixed item budgets (120 geometry, 3840 physics) rather than scaling with the core
 count, because those queues bound per-item memory: uncapped, a 64-core machine would have been handed
 40,960-deep physics queues. On machines with fewer than six cores this is a small <em>increase</em> in
@@ -114,8 +114,29 @@ pieces now carry the positioning section forward like every other modal syntax.<
 program machined one tool length low. The read point now falls back to the generic tool-number-keyed
 height and emits <code>SiemensToolOffset--TcdpRowMissing</code>.</li>
 <li><strong>G43.4 with an unresolvable H word activated RTCP with a zero-length tool.</strong> It now reports
-<code>Comp-ToolHeight--001</code> as a warning and keeps processing the block. An absent H stays silent —
-re-activating G43.4 on the modal offset id is legitimate input.</li>
+<code>Comp-ToolHeight--001</code> as a warning and keeps processing the block. An omitted H is no longer a
+silent zero: it takes the equipped tool's own offset row first (HardNc parity), then a non-zero
+modal id, and only then warns <code>Comp-ToolHeight--NoToolForOmittedH</code> (plain G43) or
+<code>Comp-ToolHeight--NoToolForOmittedHRtcp</code> (G43.4). A vacant table row warns
+<code>Comp-ToolHeight--RowMissing</code> instead of compensating a zero length.</li>
+<li><strong><code>G43.4</code> survived <code>M02</code> / <code>M30</code> and a rotary-only <code>G53</code> still coupled through RTCP.</strong> A sequence
+that ended one program with <code>M02</code> and started the next with a rotary rapid — without a <code>G49</code> —
+kept tool-centre-point control on, pinned the tip, and dragged machine XYZ through the stock. On
+SoftNc the program-end block is a <strong>reset edge</strong>: tool-length compensation (G43 / G44 / G43.4,
+Siemens <code>TRAORI</code>, Heidenhain <code>M128</code>), tilt (G68 / G68.2 / CYCLE800 / <code>PLANE</code> / TRANS), G41/G42 and
+G81–G89 cancel on the <em>next</em> block, written as explicit cancel sections so modal carry cannot copy
+the previous state. The <code>M02</code> block itself still runs under the modal state it executed with, so a
+retract on the same line as <code>M30</code> still carries tool length. A rotary-only <code>G53</code> under active
+RTCP now holds the linear axes at the previous machine position and stays McLinear, which is what
+G53 does on a control (no compensation). <strong>HardNc still carries G43.4 across <code>M02</code></strong> — the two
+engines are deliberately divergent on that edge; 3.2's default is SoftNc. A tool change that
+leaves <code>G43.4 Hm</code> alive for a different tool warns
+<code>Comp-ToolHeight--RtcpOutlivesToolChange</code> on the change block.</li>
+<li><strong>Fanuc <code>G54.1 Pn</code> (also written <code>G54 Pn</code>) was not resolved.</strong> Additional work coordinate systems
+now resolve against the extended table; a selected row nobody has entered reports
+<code>Coord-WorkOffset--AdditionalZero</code> and runs on the machine origin. <code>G59.1</code>–<code>G59.9</code> have a
+provider on the ISO presets. See
+<a href="../../anatomy/general-setup/controller/program-data/work-coordinates.html">Work Coordinates</a>.</li>
 <li><strong>Heidenhain DIN/ISO arc centres.</strong> <code>I</code> / <code>J</code> / <code>K</code> are <strong>absolute</strong> circle centres on Heidenhain —
 the ISO face of the klartext <code>CC</code> pole — not start-to-centre offsets. Reading them incrementally
 turned arcs into near-full phantom circles. Fixed in all three engines: the HardNc reader
@@ -132,10 +153,19 @@ to the workpiece over the step duration. On XYZ moves, CL files and RTCP with th
 offset the two agree (the recomputation moves forces by at most about 1e-4 relative, from the
 <code>TimeSpan</code>-quantised step duration); they part under RTCP with a tool-length offset that does not
 describe the equipped tool, on rotary-axis-limited simultaneous five-axis blocks (the tip is slower
-than F, up to a few percent), and wherever the commanded value was stale. One such stale case is
-fixed alongside: a feed block that repeats the current position (<code>X.. Y.. Z.. F1600.</code> right after
-the same point at <code>F800.</code>) lost its F word, so every following block showed and cut at the old
-feed; its <code>ActFeedrate</code> now lands.</li>
+than F, up to a few percent), and wherever the commanded value was stale.</li>
+<li><strong>A feed block that does not move lost its F word.</strong> A block that repeats the current position
+(<code>X.. Y.. Z.. F1600.</code> right after the same point at <code>F800.</code>, a common CAM habit) returned on
+zero distance before emitting anything, so every following block showed and cut at the old
+feed. Its <code>ActFeedrate</code> now lands on a zero-length linear, arc or polar block alike. The same
+stale commanded value is one of the cases the tip-feedrate split above parts on.</li>
+<li><strong>A rotary-bearing act mixed tip travel with the commanded CL feed resolution.</strong> On a pure RTCP
+rapid swing the CL point stands still, the commanded feed is ~0, and dividing the equipped tool's
+tip sweep by that resolution saturated the step count (the play appeared to hang on one line). A
+rotary-bearing act now splits by posture change (each rotary-axis delta over the rotary
+resolution) and by spindle cycles when the spindle turns — never by tip travel over the CL-feed
+resolution. Step counts of existing five-axis projects therefore shift on non-RTCP rotary
+blocks; XYZ-only programs are unaffected.</li>
 <li><strong>A <code>K0</code> word on a HardNc G02/G03 saturated the turn count.</strong> Under the default G17 plane a
 written-but-zero plane-normal word divided the axial travel by zero, the additional-turn count
 saturated to <code>int.MaxValue</code>, and one arc block became a spiral act of roughly 302 simulated years —
@@ -169,8 +169,16 @@ converting after a switch to a CL device; and
 <a class="xref" href="../../api/Hi.NcMech.Workpieces.WorkpieceService.html#Hi_NcMech_Workpieces_WorkpieceService_ClearIdealGeomCache_">ClearIdealGeomCache</a>, so loading another project stops
 rendering the previous project's target geometry.</li>
 <li><strong>Project-file operations are serialized through a zero-wait gate.</strong> A New / Load / Save / Reload
-arriving while another is in progress throws <a class="xref" href="../../api/Hi.MachiningProcs.ProjectFileBusyException.html">ProjectFileBusyException</a>
-immediately instead of racing into a file-in-use <code>IOException</code>.</li>
+/ Close arriving while another is in progress throws
+<a class="xref" href="../../api/Hi.MachiningProcs.ProjectFileBusyException.html">ProjectFileBusyException</a> immediately instead of racing into a file-in-use
+<code>IOException</code>. Close was the one operation left outside the gate, so a Close arriving during a Save
+could let the Save finish without writing and still return 200; it now refuses with the same 409
+the other five ops already used.</li>
+<li><strong><a class="xref" href="../../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_LoadProject_">LoadProject</a> and
+<a class="xref" href="../../api/Hi.MachiningProcs.LocalProjectService.html#Hi_MachiningProcs_LocalProjectService_ReloadProject_">ReloadProject</a> take a message sink</strong> for load-time
+diagnostics (a referenced STL missing on disk, a child XML that will not deserialize). A load-time
+diagnostic does not fail the load; passing no sink leaves the application log as the only witness.
+See <a class="xref" href="../../technique/api-foundations/message-management.html">Message Management</a>.</li>
 <li><strong><a class="xref" href="../../api/Hi.Machining.MachiningEquipmentUtils.MachiningEquipmentUtil.html#Hi_Machining_MachiningEquipmentUtils_MachiningEquipmentUtil_AlignWorkpieceProgramZeroToIso_">AlignWorkpieceProgramZeroToIso</a>
 computes in the machine-zero state.</strong> It reflects the assembly and zeroes every dynamic axis before
 querying displacements, so the alignment is correct even when the live machine's axes are
@@ -109,8 +109,9 @@ own, so the set is quoted as <code>3.1.175.&lt;patch&gt;</code> after its HiNc p
 it publish under their own frozen builds — looking on the feed for every package at <code>3.1.175.x</code>
 finds only HiNc. A reference left on that set receives correctness fixes for it and none of the
 capability on this page. The 3.1 builds above 3.1.175 were never published as a release set —
-which is why the release note carries one 3.2 entry where it might have carried a dozen: for a
-caller moving off 3.1.175, they were never separate releases.</li>
+which is why <strong>3.2.7</strong> is the first numbered 3.2 drop and absorbs what a caller moving off
+3.1.175 has not seen as a published set. Later 3.2 builds are incremental, the same way
+3.1.175 / 3.1.173 / … were.</li>
 </ul>

 </article>