Rotating a selection around the raw bounding-box centre moved
grid-aligned elements off the grid permanently. sceneBoundingRect()
comes from font metrics and pen widths, so the centre is almost never
a round number: with a pivot of (132.67, 101.11), an element sitting
at x=100 landed at x=133.78, and no further rotation brought it back.
Positions are written with QString::number() (%.6g), which hides the
floating-point noise but keeps the offset, so the diagram ends up
subtly misaligned with no way to repair it from the UI.
Snap the pivot with Diagram::snapToGrid(), which also follows the
user's configured X/Y grid rather than assuming the 10 px default.
Also compute the rotated offset exactly for multiples of 90 degrees
instead of going through qCos()/qSin(). The rotate actions only ever
pass right angles, and qCos(90 deg) is 6.12e-17 rather than 0, so the
trig path added error for no benefit -- four 90 degree steps did not
return a point to where it started. A quadrant is an axis swap, which
is exact; trig is kept as the fallback for any other angle.
With both, four 90 degree rotations of a grid-aligned element return
it exactly to its original position and every intermediate step stays
on the grid.
Reported by plc-user, who hit the same problem rotating graphical
primitives in the Element Editor -- discussion #618.
updateInformations() runs on every selection change and on every undo
stack index change. It unconditionally called clearToolsDock(), which
removes and hides the current editor widget, and then re-inserted the
same widget into the stack. When the editor for the new selection is
the one already shown, that means removing, hiding, reparenting and
re-adding a widget only to end up in the same state.
Look up the editor first and only clear and re-insert the tools dock
when a different editor is needed. setPart()/setParts() still updates
the contents in every case, so the visible result is unchanged.
QetLogger (discussion #644, steps 1-3) captures whatever an explicit
qDebug()/qInfo()/qWarning() call already decided to report. Most of a
session -- painting, dragging, a slow synchronous operation -- produces
no log output at all, so a silent multi-second gap in the log is
indistinguishable from the user simply not doing anything. That gap
came up directly: investigating a user-reported "the program lagged"
required inferring stalls from timestamp gaps between unrelated log
lines, which can't tell a real freeze apart from normal idle time.
EventLoopWatchdog closes that gap directly instead of inferring it. A
QTimer::PreciseTimer repeating tick (every 50ms) measures the *actual*
elapsed time since the previous tick via QElapsedTimer (monotonic,
unaffected by system clock/NTP adjustments). Qt does not queue up
missed fires for a normal repeating timer, so if the main thread is
blocked for 600ms, the timer fires once as soon as the loop frees up,
with ~600ms measured since the last tick -- that gap is the stall,
measured at its source. Only logs (via the existing qWarning() path,
so it reuses QetLogger's file/ring/rotation with no new plumbing) when
a tick is late by more than 200ms, so a healthy session produces zero
output from this class, in keeping with QetLogger's bounded-log design.
Same QET_WATCHDOG_DISABLE=1 escape-hatch convention as QetLogger's own
QET_LOG_DISABLE=1.
Deliberately not included: attributing a stall to what caused it. This
tells you a stall happened and how long -- pairing that timestamp with
gdb attached to a running session (as used for the CLI hang, PR #661)
is still how you get from "it stalled" to a root cause.
Stacked on #647 (feature-diagnostic-logging-crash) for QetLogger/
qWarning() plumbing this depends on -- diff includes its commits until
that merges.
Verified against the compiled binary, not just read: temporarily
injected a QThread::msleep(600) via a one-shot QTimer 2s after
startup, confirmed the exact expected warning
("EventLoopWatchdog: main thread stalled for 620 ms") at the right
severity through the real qWarning()/QetLogger path, then removed the
test hook and reconfirmed a normal run produces no output from this
class at all.
Stacked on the steps 1-3 branch (feature-diagnostic-logging, PR #646).
Kept as its own PR rather than folded into that one, matching the
discussion's own framing: step 4 is explicitly "the highest-risk piece
... lands last, behind its own switch."
## Step 4 -- crash-time ring flush (CrashHandler)
Installs a handler for SIGSEGV/SIGABRT/SIGBUS/SIGFPE/SIGILL (POSIX) /
SetUnhandledExceptionFilter (Windows) that flushes the in-memory ring to
a fixed crash_dump.log before the process dies.
This required reworking LogRing (step 3) to be genuinely lock-free, not
just mutex-protected: a signal handler that blocks on a lock the
crashing thread (or another thread) already holds turns a clean crash
into a hang -- no ring dump *and* no core dump, worse than doing
nothing. append() now claims a slot with a single atomic fetch-add;
dumpToFd() reads the preallocated entries directly and writes them with
write(2) only, looping on EINTR/short writes. Accepted tradeoff: at most
one entry can be read torn if a crash lands mid-append into that exact
slot -- documented in logring.h, and the alternative (a seqlock to
detect and retry) wasn't judged worth the complexity for that window.
Other invariants implemented per the discussion:
- sigaltstack with a static 64 KiB buffer, SA_ONSTACK -- a stack-
overflow SIGSEGV has no usable stack for a handler without one.
- Nothing under the actual handler touches Qt, QString or the
allocator: the dump path and a small header (version/git/OS/Qt) are
precomputed into fixed char buffers by install(), which runs once at
startup in normal context.
- Atomic test-and-set so only the first crash writes a dump; a second
concurrent/nested fault goes straight to restore-and-re-raise.
- After writing, the handler restores SIG_DFL and re-raises (POSIX) /
returns EXCEPTION_CONTINUE_SEARCH (Windows) so the OS's own crash
path -- core dump, Windows Error Reporting -- still runs. A handler
that "fixed" the crash by swallowing the signal would destroy exactly
the post-mortem evidence this whole design exists to preserve.
Tested in this environment: POSIX/Linux only, all five signals. Sent
each directly to a running process and confirmed (a) crash_dump.log is
written with the correct header and ring contents, mode 0600, and (b)
the process still terminates via the signal with the kernel's own
"core dumped" flag set (exit code 128+signal, confirmed for all five).
The Windows path is implemented per the discussion's guidance but is
untested -- no Windows build available in this sandbox.
## Step 5 -- getting the data back out
- QETApp::checkCrashDump(), called from checkBackupFiles() only when
there's no stale project file to recover this run (so the two
prompts never both show, per the discussion), offers an unretrieved
crash dump via DiagnosticsReportDialog and then deletes it regardless
of the user's choice -- offered exactly once.
- A new "Aide > Enregistrer un rapport de diagnostic..." action
(QETMainWindow) builds the same kind of report from the *current*
session (QetLogger::buildDiagnosticsReport(): header + this session's
log file) for a manual "attach this to a bug report" flow, not tied
to a crash.
- Both go through QetLogger::redact() before ever reaching the user:
the one redaction implemented is a literal replace of the home
directory with "~", since an absolute path under it leaks the
account name. The discussion's fancier "optionally redact project
filenames too" isn't attempted -- reliably telling a project path
apart from arbitrary log text is a much fuzzier problem than a
literal prefix match.
- DiagnosticsReportDialog shows the full (already-redacted) content
before saving, per the discussion: "the user is about to attach this
to a public tracker."
Verified in a real GUI session (Xvfb): triggered a SIGSEGV, relaunched,
confirmed the crash-report dialog appears with the right header/content,
confirmed it does not reappear on a second relaunch, and confirmed the
manual "Save report" action produces a correctly-formatted report and
saves it to a chosen path.
Built clean, no new warnings.
## Build systems
Registered in both: cmake/qet_compilation_vars.cmake, and
qelectrotech.pro. The .pro needed explicit globs for the new
sources/logging/ui/ subfolder -- sources/logging/*.{h,cpp} was already
globbed, but unlike the other ui/ subfolders that one had no entry of
its own, so diagnosticsreportdialog.{h,cpp} would not have been built
under qmake.
The job trigger was narrowed to push:tags a while back, but the
job-level 'if: github.ref == refs/heads/master' was left in place.
A tag push never has github.ref == refs/heads/master, so the two
conditions are mutually exclusive: the job trigger fires only on
tag pushes, while the guard only allows master-branch refs, meaning
the job has been silently skipped on every run since the trigger
was narrowed.
This is also the likely source of the recent Git LFS bandwidth/
storage overage: while the trigger was still push-to-master (pre-
narrowing), this job ran on nearly every commit and committed a
new QElectroTech.qch (LFS-tracked) each time, via the auto-generated
update-qch PR -- accumulating one LFS object version per run."
ProjectView::initWidgets() called insertSpacing(1, 10) on a QHBoxLayout
that was still empty, inserting past the end of the item list. The
corrupted layout crashed later in QWidget::setLayout() via
QLayoutPrivate::reparentChildWidgets() and QBoxLayout::itemAt().
Use addSpacing(10) instead, which is equivalent for an empty layout.
BorderTitleBlock::slot_setAutoPageNum was removed in 471f876 ("Remove unused signal", 2023-10-17) without noticing autonumberingdockwidget.cpp still referenced it via old-style
SIGNAL()/SLOT() macros, which fail silently at runtime instead of producing a compile error. This has remained broken on master ever since; a fix (57572a2, "Fix two broken signal
connections") exists on the unmerged qt6-cmake-elevatormind-merged branch but that one only commented the lines out without solving the underlying issue.
Removed the dead code entirely and call BorderTitleBlock::importTitleBlock() directly on the active diagram in on_m_folio_cb_activated() instead. The same mechanism is already
used elsewhere in the codebase (undo command, new-diagram creation, XML loading) to push TitleBlockProperties into a diagram and trigger a folio numbering recompute via needFolioData().
27dcd5e renamed BorderTitleBlock::diagramTitleChanged to informationChanged and updated diagram.cpp accordingly, but diagramview.cpp still used the old string-based SIGNAL()/SLOT()
macro referencing the removed signal name, which failed silently at runtime instead of at compile time. The diagram/window title never refreshed after title block changes.
Observed when going through the warnings.
Updated the connect to use informationChanged with modern pointer-to-member syntax, matching diagram.cpp's own connect to the same signal.
removed in Qt6, replaced by mappedInt/mappedWidget/mappedString.
What was broken:
* the logo-conflict rename dialog
* the system tray show/hide toggle
* the Window menu
* export dialog's per-diagram preview controls
Switched to the modern mappedInt/mappedWidget signals with pointer-to-member connect(), guarded for Qt < 5.15 until Qt5 can be dropped.
connect() used the string-based currentIndexChanged(QString) signal, which was removed from QComboBox in Qt6. Selecting a conductor/element/folio auto-numbering context in the
combo box in the project properties dialog never updated the other fields.
Switch to currentTextChanged with the modern pointer-to-member connect() syntax, which also catches signal/slot mismatches at compile time. This is still backward compatible with Qt5 (as long as this is needed).
Never leave a collection folder without a name (replaces #622)
Mark elements-folder with an exclamation-mark, when "qet_directory" is missing or faulty.
The unconditional early return was narrowed to non-directories only, so
that the just-added warning badge could be picked up once setUpData()
resolved m_qet_directory_unreadable asynchronously. But every directory
then called setIcon() on every single data(Qt::DecorationRole) query --
not just once -- and QStandardItem::setIcon() -> setData() emits
dataChanged() unconditionally (QIcon has no equality check to suppress
it). QTreeView handles dataChanged() by recomputing the row's size hint,
which re-enters data() for the same index, calling setIcon() again:
unbounded mutual recursion, confirmed by an isolated reproduction to
overflow the stack in a single frame (100k+ frames) well before the
first paint completes. Matches plc-user's report of a segfault right as
the elements tree begins drawing.
The race the guard was widened for doesn't actually occur:
ElementsCollectionModel only attaches itself to the tree view (the only
way data() becomes reachable) from loadingFinished(), which fires after
the QtConcurrent::map over every item -- this one included -- has
already finished. m_qet_directory_unreadable is therefore always final
before setUpIcon() can run for the first time, so the plain, always-only-
once guard is sufficient and the badge still works correctly.
`qelectrotech --resave examples/schema_indus.qet out.qet` never returns.
It is not slow -- ten minutes of wall clock consumed 0.16s of CPU, so it
is blocked, not working. The GUI opens the same project without
complaint, so the file is fine and the fault is in the headless path.
A backtrace of the stuck process:
main
CLIExport::run
QETProject::QETProject(QString const&, QObject*)
QETProject::openFile(QFile*)
QETProject::readProjectXml(QDomDocument&)
QET::QetMessageBox::warning(...)
QDialog::exec() <- waits forever
That project records version="0.3", so loading it raises the "partially
compatible with your version" warning. Interactively somebody presses
Open; with no display nobody can, and exec() spins its event loop
indefinitely. Any modal box reachable while loading does this -- the
version warning is just the one an example file happens to trigger.
Fixed at the wrapper all 52 call sites already go through rather than at
the one warning, so the whole class is closed: QetMessageBox gains a
non-interactive mode which writes the message to stderr and returns an
answer instead of constructing a dialog. main.cpp turns it on in the
CLI branch, beside the existing setBackupEnabled(false).
The answer is the caller's defaultButton when it gave one, otherwise the
first "carry on" button offered (Ok, Open, Yes, Save...), otherwise the
first button set. Both warnings in readProjectXml offer Open|Cancel and
abort on Cancel, so they resolve to Open and the project loads, which is
what a batch invocation wants. The text still reaches the user on
stderr, where previously it was lost inside an invisible dialog.
GUI behaviour is unchanged: the flag defaults to false and is set in
exactly one place, the command-line branch of main().
Verified: schema_indus.qet goes from hanging to resaving in 0.3s; all 23
example projects now complete a double-resave with element, conductor,
terminal and uuid sets intact; unit tests pass.
RotateSelectionCommand's existing "Pivoter" action (Space) only ever
bumps each selected item's own rotation property -- QGraphicsItem's
setRotation() spins an item around its own local origin and never
touches pos(). Select three elements arranged in a row and rotate:
each spins 90 degrees individually, but the row stays a row. That's
"rotate each item," not "rotate the group."
Add a rotate_as_group parameter to RotateSelectionCommand (default
false, so the existing action and its one call site are unchanged).
When set, it computes a shared pivot once -- the bounding-box center
of the whole selection -- and queues a second, parallel "pos"
QPropertyUndoCommand alongside the existing "rotation" one, rotating
each item's position around that pivot by the same angle.
Scoped the position change to Element/IndependentTextItem/
DiagramImageItem only: these are the only selectable types with
scene-space pos(). ConductorTextItem, DynamicElementTextItem and
ElementTextItemGroup are all parent-relative children (confirmed by
reading their constructors), so when their owning Element is also
selected and gets its own pos() rotated, they're carried along for
free by Qt's normal parent/child transform propagation -- exactly
what the existing "skip rotation if parent is also selected" guard
already assumes for those three cases.
Exposed as a new, separate action ("Pivoter le groupe", Shift+Space)
next to the existing one rather than changing Space's behavior, since
some workflows may rely on the current per-item rotation.
"Add folio" always appends to the end of the project, ignoring
whatever folio is currently selected in the left panel -- even though
the panel already tracks the selected diagram's position for its
existing move up/down/top actions, and QETProject::addNewDiagram(pos)
already accepts an arbitrary insertion index, pushed as an undoable
AddDiagramCommand (QetGraphicsTableFactory::create() already relies on
this exact mechanism to insert a folio right after a specific one).
Add two new context-menu actions that compute the target position from
the selected diagram's folioIndex() and pass it straight through the
existing machinery -- no changes needed to QETProject or
AddDiagramCommand. New requestForNewDiagramAt/addDiagramToProjectAt
signal/slot pair added alongside the existing
requestForNewDiagram/addDiagramToProject rather than changing it, so
the plain "Add folio" action's append-at-end behavior is untouched.
Drawing tools on the diagram canvas always started new shapes and free
text from a fixed hardcoded default (Qt's plain QPen()/QBrush(), and
the static Preferences font) -- changing a shape's color or a text's
font had zero effect on what the next new item of that type got, even
within the same editing session.
Add LastUsedStyle: a small in-memory, session-scoped static helper
(no QSettings, no persistence across restarts -- this is a live "what
did I just use" value, not a new app-wide default). Write side hooks
capture the value right where the properties editors already apply a
change (ShapeGraphicsItemPropertiesWidget::associatedUndo(), both the
live-edit dock path and the modal editProperty() dialog path; and
IndiTextPropertiesWidget::on_m_font_pb_clicked() right after the font
dialog returns). Read side hooks apply the stored value, if any, to a
newly created item: DiagramEventAddShape::mousePressEvent for shapes,
IndependentTextItem's constructor for free text (falling back to the
existing QETApp::indiTextsItemFont() Preferences default otherwise).
Doesn't touch the element/symbol editor's own drawing tools (a
separate subsystem) or add last-used text color (no color control
exists in the text properties UI to originate it from yet).
@plc-user asked (review on #633) for a way to see a broken folder
directly in the tree instead of only on tooltip hover, originally
suggesting a "FixMe: " text prefix on the displayed name. That name is
reused verbatim in dialog titles and status-bar messages elsewhere
(elementscollectionwidget.cpp), so baking a prefix into it would leak
into those too. An icon badge gets the same visibility without
touching the name value.
setUpIcon() overlays a small warning glyph on the folder icon when
m_qet_directory_unreadable is set. Also drop the "already has an icon,
skip" guard for directories specifically: that flag is only known once
the async setUpData()/localName() job completes (QtConcurrent::map), so
without this a directory painted before that finished would have its
plain folder icon cached forever and never pick up the badge.
Implements steps 1-3 of discussion #644 (deliberately not steps 4/5 --
no signal handler / crash flush, no diagnostics UI; see below).
## Step 1 -- fix the existing logger (bugs, no new behavior)
- One QFile handle held open for the whole session under a mutex,
instead of opening and closing the log file on every single message.
- The log directory and the session's date-stamped filename are
resolved exactly once, in the new QetLogger::init() called explicitly
from main() immediately before qInstallMessageHandler() -- not
recomputed per message, so a session that runs past midnight now
stays in one file instead of silently splitting.
- Age-based retention now uses lastModified() instead of lastRead():
opening a log to attach it to a bug report no longer resets its
retention clock.
- stderr and file output both encode UTF-8 explicitly (toUtf8()),
replacing stderr's toLocal8Bit() and the file stream's previously
Qt5/Qt6-inconsistent default encoding.
## Step 2 -- size-capped rotation + hardening
- The previously-unbounded daily file is now capped at 2 MiB and
rotated (kMaxFileBytes/kRotationKeep in QetLogger), keeping
<date>.log plus <date>.1.log .. <date>.4.log; oldest is dropped.
- Each message is truncated to 4 KB with a "...[truncated N bytes]"
marker before it reaches the ring or the file.
- Control characters (newlines, tabs, other non-printables) in message
content are escaped, since much of what QET logs is externally
controlled (file paths, element names, font strings out of a .qet
file) -- left unescaped, an embedded '\n' could forge log lines.
- The log file is refused if a symlink already exists at that path,
and is created/rotated owner-read/write only.
## Step 3 -- in-memory ring buffer
- LogRing (sources/logging/logring.h) is a fixed-capacity, always-on
ring of the last 4096 log lines, preallocated once at construction
(4096 * 512 B = 2 MiB) so append() never allocates. Entries are
stored as plain pre-formatted bytes in fixed-size slots -- the shape
discussion #644 specifies so a *future* crash handler could dump it
with nothing but write(2), even though no such handler exists yet.
Thread-safe via a plain QMutex (the lock-free requirement in the
discussion applies specifically to a signal-handler read path, which
this step doesn't add).
## Escape hatch
QET_LOG_DISABLE=1 in the environment at startup bypasses all of the
above -- no ring, no file, no rotation -- falling back to a minimal,
self-contained stderr passthrough that doesn't share any code with the
new formatting/sanitization path, so it stays usable even if that path
is what's misbehaving.
## Deliberately not included (per the discussion's own phasing)
- No signal handler / crash-time ring flush (step 4) -- the discussion
flags this as the highest-risk piece, explicitly meant to land last
and behind its own switch once the rest is proven.
- No diagnostics export UI (step 5).
- No log categories, session header, repeat collapsing or rate
limiting -- listed under "best practices worth building in", not
part of steps 1-3.
## Testing
Built clean, no new warnings.
Verified with real runs (QT_QPA_PLATFORM=offscreen, isolated HOME):
- Log file created at the expected dataDir()/YYYYMMDD.log path, mode
0600.
- A full startup's worth of real messages (translations, MachineInfo's
system dump, collection loading) written correctly; every one of the
231 lines in one run starts with a proper timestamp -- confirmed the
sanitizer correctly escapes the raw embedded newlines/tabs in
MachineInfo's multi-line CPU/GPU description fields into visible
\n/\t sequences rather than letting them fragment the log.
- QET_LOG_DISABLE=1: zero log files created, stderr still worked via
the independent legacy path.
- Rotation: pre-filled a log to just under the 2 MiB cap, ran a normal
session, confirmed it rotated to <date>.1.log (still 0600) with a
byte-clean split (no truncated/duplicated line at the boundary) and
a fresh <date>.log picked up from the next line.
Placing an auto-numbered element or conductor advances a shared
NumerotationContext counter (QETProject::addConductorAutoNum/
addElementAutoNum) as a side effect that sat entirely outside the undo
stack. Undoing the placement removed the visible number but left the
counter advanced, so every undo of an auto-numbered placement silently
burned a number, with no way to get it back short of a manual reset.
Adds SetAutoNumContextCommand, a small QUndoCommand storing the old/new
NumerotationContext and calling the matching add*AutoNum() setter on
undo()/redo() -- the same shape QPropertyUndoCommand already uses next
to it in ConductorAutoNumerotation::applyText().
Wires it into the two conductor call sites (the static newProperties(),
and numerateNewConductor(), both in ConductorAutoNumerotation) and the
element call site (Element::setUpFormula(), called from
DiagramEventAddElement::addElement() when a new element is dropped onto
a diagram). setUpFormula() now takes an optional parent QUndoCommand;
addElement() calls it before pushing its own undo_object so the counter
change lands in the same undo macro as the element's placement -- one
Ctrl+Z reverts both together, instead of leaving the counter adrift.
The project-properties config dialog's own add*AutoNum() calls (editing
the numbering rule itself, not a side effect of placing something) are
deliberately left untouched, as are the load-time folio-sequential
bookkeeping calls in Diagram::loadElmtFolioSeq()/loadCndFolioSeq() and
the bulk folio-renumbering passes in QETProject -- none of those run as
part of an undoable user gesture.
Implements the scope proposed in discussion #608.
ElementInfoWidget's fixed ~40 predefined ELMT_* keys had no way for a
user to add a genuinely new element-info key, even though DiagramContext
already stores/round-trips arbitrary keys generically via toXml()/fromXml().
Adds an "Ajouter une propriété personnalisée" button that appends a
CustomElementInfoPartWidget row (both key and value user-editable,
unlike the fixed ElementInfoPartWidget rows bound to one predefined
key). The typed key is validated live against the existing
DiagramContext::isKeyAcceptable() and flagged with a red border when
it doesn't match, instead of silently dropping it. Any key already
present on the element that isn't one of the predefined/special keys
is re-displayed as a custom row on next selection.
Implements the scope proposed in discussion #611.
Suggests the element's filename (without its .elmt extension) as the
default save name when exporting to SVG, instead of only defaulting to
the customElementsDir with no filename. Addresses plc-user's review
suggestion on PR #637.
Implements discussion #605. The diagram editor can already export a folio
to SVG; the element editor, where a single .elmt symbol is drawn, had no
export capability at all -- confirmed by grepping its header for "export"
before starting: nothing.
## Renders the live scene, not ElementPictureFactory's cache
The discussion proposed sourcing this from ElementPictureFactory's cached
per-element QPicture (m_pictures_H), the one used for the elements-panel
preview icons. Checked that cache's actual invalidation before building on
it: nothing in the editor ever tells it to drop an entry on edit, and it is
keyed by the element's on-disk uuid. So for any element already previewed
once in the panel, exporting from the cache would silently produce stale
content after any edit; for a brand-new, never-saved element, no entry
would exist at all. Neither is acceptable for a File > Export action a
user expects to reflect what's on screen right now.
Renders ElementScene directly instead, the same way
ExportDialog::generateSvg() already renders the live Diagram for the
diagram editor's own SVG export: no new drawing logic, only a new playback
target (QSvgGenerator instead of the screen), sized to the element's own
content bounds via the existing elementSceneGeometricRect() helper.
## Hotspot cross excluded from the export
ElementScene::drawForeground() draws the red origin/hotspot cross on every
render() call, unconditionally -- it's an editing aid, not part of the
element being drawn, and diagram editor's SVG export has no equivalent
problem since Diagram doesn't draw one. Added a settable
hotspotVisible flag, defaulting to true (the existing editing view is
completely unaffected) and turned off only for the duration of the export
render() call.
## Verified end-to-end via a real Xvfb session, not just a build
Opened a real shipped element (en_60617_05_06_04.elmt, "Phototransistor"),
exported it, and rendered the resulting SVG back to a bitmap with a small
QSvgRenderer-based harness -- pixel-identical in shape to the element as
shown in the editor. Confirmed the file is valid XML and contains no
red/#ff0000 stroke (the hotspot cross did not leak in).
Then the case the whole "render live, not cached" decision was about:
opened the same element, drew a new line with the line tool, and exported
again *without saving*. The new line is present in the exported SVG.
git status on the source .elmt file after both exports shows it completely
untouched -- the export is read-only and reflects live, unsaved editor
state, exactly the property a cache-based implementation would have gotten
wrong.
Built clean, no new warnings.
Discussion #599's own scope explicitly deferred this ("Related, not
proposed here... a natural follow-up once basic pan/zoom motion works").
Basic pan/zoom now works (previous commits on this branch), so this adds
it -- generically, for whichever SpaceMouseBackend is in use, not tied to
libspnav specifically, matching the seam the previous commit built.
## Reuses ShortcutManager instead of inventing a second action registry
ShortcutManager is already an app-wide registry of every named, rebindable
action -- undo, redo, rotate selection, cut/copy/paste, autonum configure,
and dozens more -- each carried by a live QAction or QAbstractButton. A
device button binding to one of *those* ids, rather than to a bespoke
QET-3D-mouse-only action list, means the discussion's own examples
(rotate/mirror/undo) are available for free, and any action added to the
app in the future is automatically bindable too.
Added ShortcutManager::trigger(id): find the first still-alive target for
an id and call QAction::trigger() or QAbstractButton::click(), whichever
it is. Deliberately not disambiguated by which window is currently active,
unlike SpaceMouseListener's own pan/zoom dispatch -- a target's owning
top-level window isn't reliably discoverable from a bare QAction. Correct
in the overwhelming common case of one open editor window; documented in
the header as a known simplification, not silently assumed correct.
## The binding itself: SpaceMouseButtonMap
A thin QSettings-backed button-number -> action-id map, unbound by default
for every button on every device -- nothing happens on any button press
until the user opens Configuration > Souris 3D and binds something,
matching this whole feature's "silent until asked for" default.
## Backend side: SpaceMouseBackend::buttonPressed(int)
Added to the platform interface alongside the existing motion() signal.
SpnavBackend now handles SPNAV_EVENT_BUTTON (previously explicitly
ignored) and emits on press only -- release is not reported, since nothing
downstream has a use for it. A future non-spnav backend implements the
same signal and gets button support for free through
SpaceMouseListener::applyButton(), without that logic being duplicated or
re-verified per backend -- the same reasoning the previous commit's seam
was built around.
## Configuration UI: SpaceMouseConfigPage
Modelled directly on the existing ShortcutsConfigPage -- same QTableWidget
shape, same "persist on applyConf(), not live" contract -- one row per
binding: button number (spin box, unbounded, since button count and
numbering genuinely vary from 2 to 30+ across real devices and this could
not be checked against hardware) and action (combo box populated from
ShortcutManager::instance().allShortcuts(), the exact same live registry
the Shortcuts page itself lists). Only added to the Configuration dialog
when QET_SPACEMOUSE_SUPPORT is compiled in.
## Verified, including the one thing that doesn't need hardware to prove
Rebuilt from scratch both ways: option off adds zero new object code
(confirmed via a forced rebuild of the one unconditionally-changed file,
shortcutmanager.cpp, which alone picked up new warning-free code); option
on compiles all four new/changed files warning-free and links clean.
The backend's button *detection* (SPNAV_EVENT_BUTTON -> buttonPressed
signal) still cannot be verified without a real device or daemon -- same
limitation as the motion path from the previous commits, stated plainly
rather than glossed over.
What *is* fully verified, because none of it needs hardware:
- SpaceMouseButtonMap: unbound by default, set/read-back, clearing via an
empty id, enumeration -- all confirmed via a standalone harness linked
against the real compiled objects.
- ShortcutManager::trigger(): registered a real QAction, confirmed
trigger() fires it exactly once and returns true; confirmed it returns
false (not a crash) for an unknown id.
- SpnavBackend: constructs safely with no daemon present (isAvailable()
false, as it must be), and both its motion and buttonPressed signals
are correctly wired per Qt's own metaobject data (QSignalSpy).
- The configuration page end-to-end, via a real Xvfb session: opened
Configuration > Souris 3D, confirmed the action combo box lists the
live, real ShortcutManager registry (undo, rotate, cut/copy/paste,
dozens more -- not a mock), added rows, edited the button number,
removed rows, selected "Éditeur de schémas — Pivoter" (Rotate -- the
discussion's own example) for button 3, clicked OK, and confirmed via
the actual settings file that it persisted exactly as
"buttons\3=diagrameditor.rotate_selection". Reopened the dialog and
confirmed it read back correctly. This is a full, real round trip
through the UI, not a claim.
The user asked for phase 2 (Windows/macOS via 3Dconnexion's proprietary
3DxWare SDK) on top of #635. This sandbox has no 3DxWare SDK, no Windows
toolchain, and no macOS toolchain -- nothing to compile, link, or run a
single line of platform code against, unlike the Linux/libspnav path,
which was built and actually tested here for real. Writing 3DxWare
integration code that has never even built would be a materially weaker,
unverifiable thing sitting in this PR, so it is not in this commit.
What is: the structural seam that makes adding it later a contained,
reviewable change instead of a rewrite of code that already works.
## Before
SpaceMouseListener did three unrelated things in one class: own the
libspnav connection, read spnav events, and apply motion to the active
DiagramView. A Windows/macOS backend would have had to either duplicate
all of the DiagramView-facing logic (the pan/zoom calls, the Z-to-zoom-
factor mapping, the "which view is active" lookup -- all already verified)
or bolt onto the same class with a maze of #ifdefs. Either way, touching
that file again would put the already-tested Linux path back in scope for
review.
## After
- SpaceMouseBackend: a tiny interface (isAvailable(), a motion(dx,dy,dz)
signal). A backend's only job is owning one platform's connection to the
driver/daemon and translating its native event into this one signal.
- SpnavBackend: the libspnav code from the previous commit, moved behind
that interface with no behaviour change -- still spnav_open() in the
constructor, still a QSocketNotifier on spnav_fd(), still silent when no
daemon/device is present.
- SpaceMouseListener: now backend-agnostic. Owns whichever backend the
platform provides, applies its motion to the active DiagramView exactly
as before. The DiagramView-facing code (pan/zoom calls,
zoomFactorForZAxis) did not need to change at all -- only its input
changed from a spnav_event_motion struct to three plain ints.
A future 3DxWare backend implements SpaceMouseBackend, is selected in
SpaceMouseListener's constructor behind its own
QET_SPACEMOUSE_BACKEND_3DXWARE guard (see the comment marking exactly
where), and never has to touch SpnavBackend or SpaceMouseListener's
DiagramView-facing half.
## CMake: one user option, one define per backend
QET_ENABLE_SPACEMOUSE is unchanged as the single option a user sets.
Internally, find_spacemouse.cmake now decides *which* backend (if any)
that resolves to: on Linux with libspnav found, QET_SPACEMOUSE_BACKEND_SPNAV
plus the umbrella QET_SPACEMOUSE_SUPPORT. Turning the option on anywhere
else today downgrades cleanly with a warning naming discussion #599,
instead of trying (and failing) to find libspnav on a platform that
doesn't ship it. Adding 3DxWare later means adding one more branch here,
not restructuring this file.
## Verified this is a pure refactor, not just "still compiles"
Reconfigured and rebuilt both ways from scratch:
- option off: unchanged from before -- no new source files compiled, zero
new object code.
- option on: both new files compile with zero warnings, binary still
links against libspnav.so.0 (confirmed via ldd), and run to completion
in this environment (which has no spacenavd) with zero crashes and zero
spnav-related output -- identical to before the refactor.
- zoomFactorForZAxis re-linked and re-run in isolation: identical output
to the pre-refactor commit (z=0 -> exactly 1.0, z=+-350 -> 1.35/0.65),
confirming the math moved unchanged rather than being reimplemented.
Implements discussion #599's phase 1 (Linux, libspnav): a 3Dconnexion
6-DOF device pans and zooms the active diagram view, via spacenavd.
## Off by default, zero cost when off
QET_ENABLE_SPACEMOUSE (cmake/developer_options.cmake) is OFF. Verified in
two separate build directories from a clean configure: with it off, the
new cmake/find_spacemouse.cmake step runs and does nothing (no library
lookup, no definition, no new source files compiled), and qetapp.cpp/.h
produce zero new object code -- both are entirely #ifdef'd out. The
default build is byte-for-byte the same shape as before this commit.
With it on, libspnav is located via its pkg-config file (spnav.pc, shipped
by libspnav-dev on Debian/Ubuntu and equivalent packages elsewhere). If the
option is on but the library isn't found, this does not hard-fail
configure: it downgrades back to off with a warning, so an opt-in feature
never blocks a developer who doesn't have the library installed.
## No new navigation logic -- a new input source for the existing one
DiagramView::wheelEvent() already turns a physical wheel's delta into
horizontalScrollBar()/verticalScrollBar() calls for pan and a
zoom(1 + value/1000) call for zoom -- see diagramview.cpp:661-687.
SpaceMouseListener calls exactly those same primitives from spnav motion
events instead of wheel events. It does not reimplement panning or
zooming.
## Safe by construction even when compiled in
The overwhelming majority of users of a build with the option on still
won't have spacenavd running or a device attached -- that must never
surface as an error dialog or a startup warning. SpaceMouseListener::
isAvailable() reflects this: spnav_open() failing is treated as the
ordinary case, not an error, and the object then does nothing at all.
Verified for real in this environment, which genuinely has no spacenavd
running: built with the option on, ran the resulting binary to completion,
and confirmed zero crashes and zero spnav-related output of any kind --
the silence is the point.
Motion is read via a QSocketNotifier on spnav_fd() (event-driven, no
polling loop, no idle cost) and applied to whichever DiagramView is
currently active, found via qApp->activeWindow() -> QETDiagramEditor ->
currentProjectView() -> currentDiagram(): a 6-DOF device is one ambient
input source for the whole application, not something tied to a
particular window, so there is exactly one listener, owned by QETApp.
## What could not be verified without hardware
The Z-axis-to-zoom-factor mapping (SpaceMouseListener::zoomFactorForZAxis)
is a pure function specifically so it could be tested without a live
device: confirmed a centered device (z=0) yields exactly 1.0 (an exact
no-op, not an epsilon-off value that could trip DiagramView::zoom()'s
>=1 branch), and that push/pull produce symmetric zoom-in/out factors.
What genuinely cannot be checked in this environment: which physical axis
is "left/right" vs "up/down" vs "push/forward", their sign, and whether
the ZOOM_DIVISOR/PAN_SCALE constants feel right on real hardware. Both are
named constants specifically so recalibrating them is a one-line change
once someone with a device tries it -- flagged plainly in the PR rather
than presented as verified.
## Not in this commit
Windows/macOS (proprietary 3DxWare SDK, materially bigger lift) and
device button mapping are both explicitly out of scope for this phase per
the discussion.
ispyisail