tst_toolbarsettings, tst_diagramtoolbarsettings and tst_gesturesettings
never set an organization or application name. On macOS and Linux,
QSettings then falls back to an "Unknown Organization" file, so the
tests pass there. On Windows it has no registry key: every write is
dropped and every read gives the default, and nine checks fail.
Each test now sets its own organization and application name in
initTestCase(), as tst_scriptingsetting and tst_sheetbackgroundsetting
already do. That also keeps the tests away from the settings of the
account that runs them.
Tests only; no change to the app or its files.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Closing a title block template editor crashed it. The editor deletes
itself on close; QWidget's destructor then deletes its children, and
some of them signal the editor as they go:
- the scene deselects its cells, so the view emits
selectedCellsChanged() to the editor's slot of the same name (when a
cell was selected);
- the undo stack is cleared, so it emits cleanChanged() to
updateEditorTitle() (when the template was edited).
Both slots then run on an object whose destructor has already run. A
Debug build aborts ("Called object is not of the correct type (class
destructor may have already run)"); a release build calls a member
function of a destroyed object.
Disconnect every child of the editor from it in the editor's
destructor, before the base destructors delete the children.
No file-format change, no change while the editor is open.
Tests: none automated; the editor needs the whole application.
Checked by hand (see the PR).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
Every headless call failed with the macOS package: _launch_executable()
copied the program out of the signed .app bundle, and macOS kills such a
copy at launch (exit 137). The copy has also lost the path to the bundle's
Qt frameworks. Windows already ran the original for the same kind of
reason (DLLs beside the program); macOS now does too.
That is safe for the same reason as on Windows: the export flags and
--run return before QElectroTech constructs SingleApplication, so there
is no running instance to hand the call to. Linux keeps the private copy.
Reported and tested by DarkBiscuits on macOS 26.7 with the r10796 DMG:
with this one change and QT_QPA_PLATFORM=offscreen, every tool works
headless with no Dock icon.
test_qet_mcp: 421 tests pass on Linux against a master build with
QET_BINARY/QET_ELEMENTS/QET_EXAMPLES and scripting on, none skipped;
test_which_systems_copy fails with darwin removed from the rule.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Generated by tools/qet-en-source/regenerate.sh, not edited by hand. The
tree is the fresh conversion; this commit only joins it to the PR's
history so the branch is never force-pushed.
Generated by tools/qet-en-source/regenerate.sh, not edited by hand. The
tree is the fresh conversion; this commit only joins it to the PR's
history so the branch is never force-pushed.
"Position automatique" sorted labels as plain strings, so XAUZ1:10
came before XAUZ1:2. Add QETUtils::naturalLessThan (digit runs compared
by value, case insensitive; QCollator numeric mode is ignored by some
Qt backends) and use it in SortTerminalStripCommand, FreeTerminalModel
and the terminal strip tree dock.
Terminal elements only offered eight information fields in the
diagram editor (formula, label, comment, designation, manufacturer,
manufacturer reference, machine manufacturer reference, supplier).
That subset was chosen in faf4235be, before the size, quantity and
auxiliary fields existed, so a terminal could not be given a width,
height, depth or type.
The element editor moves the selection in fine steps with
Ctrl+arrow. The diagram editor used Alt+arrow for its fine
movement and Ctrl+arrow for aligning a text group, so the same
modifier did different things in the two editors.
Swap the two so the diagram editor matches the element editor:
- Ctrl+arrow: fine movement (unchanged step, still configurable via
diagrameditor/key_fine_Xgrid and key_fine_Ygrid)
- Alt+arrow: align the selected text group (Left/Up/Right)
Only the modifier checks change, the behaviour of each action is
untouched.
The folio grid steps, diagrameditor/Xgrid and Ygrid, are read from the
settings in six places with a plain toInt() and used as they come. The
preferences page cannot store anything below 1, but a hand-edited or
damaged settings file can hold 0, a negative number or text:
- Diagram::snapToGrid() divides by the step, so 0 is a SIGFPE on the
first click that places a symbol;
- Diagram::drawBackground() runs "while (g_x % xGrid)" and then loops
"gx += xGrid", so 0 crashes every repaint and a negative step never
ends;
- the paste, duplicate and align paths divide by it or multiply with it.
Add foliogrid.h, a header-only helper: FolioGrid::step() reads a
settings entry and returns the built-in step (Diagram::xGrid, 10) when
the entry is missing, not a number, below 1 or above what an int
holds (QVariant::toInt() wraps such a value around). Every reader of
the two keys goes through it; the settings page, which only writes the
spin box values, is unchanged.
No file-format change, and no change for any step the preferences page
can produce.
Tests: tst_foliogrid covers the helper: 1, 10, "7", int max are kept;
missing, 0, -5, "ten", "", "nan", 99999999999 (as text and as a
number) and int max + 1 fall back; 7.9 rounds to 8 as before; and the
two settings keys through a QSettings scope of the test's own. The
helper is new, so the test cannot fail on master; the six readers are
the replacements in the diff.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
Two position comparisons handed to std::sort were not strict weak
orderings, which std::sort requires; with the wrong kind of comparator
the sort is undefined behaviour (libstdc++ can read past the range,
MSVC debug builds assert "invalid comparator").
- comparPos(), used when renumbering the elements of a project, ended
with "<=" on x and y, so two elements at the same position (pasted at
the origin, placed by a script, stacked symbols) were each "before"
the other.
- The terminal numbering dialog compared positions with a 1 px
tolerance ("within 1 px counts as aligned, then compare the other
axis"), which is not transitive: terminals at x = 2, 1.1 and 0.2 give
a < b, b < c and c < a.
Move the comparisons into positionorder.h, a header-only helper:
xThenY()/yThenX() with "<", and roundedXThenY()/roundedYThenX(), which
round the positions to whole pixels first so that items a fraction of a
pixel apart still count as aligned, as the tolerance meant to, while
staying transitive. comparPos() keeps its folio and row-letter stages
and calls xThenY() for the last one.
No file-format change. Elements at distinct positions sort exactly as
before; the terminal order changes only for terminals less than a pixel
apart that straddle a half-pixel boundary.
Tests: tst_positionorder checks each order on a grid of awkward
positions by brute force (irreflexive, asymmetric, transitive, with a
transitive equivalence), the three-terminal cycle, twenty items at one
position, and that std::sort leaves the list sorted and intact. The
helper is new, so the test cannot fail on master; the call sites are
the two replacements in the diff.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
The interface text is now English at the source, so the untranslated
checkbox and tooltip read in English.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every tr() source string in the code and the forms is now the English
text, and French is a translation (qet_fr.ts) like the other languages.
Generated, not hand-edited: master ef795a21e5 converted by
qet-en-source (harness 4c4ddc9).
3144 messages get an English source, 665 already had one, 0 left French; 8 shared-wording pairs disambiguated.
C++: 2904 literals in 233 files. Forms: 809 strings in 65 files. All 34 .ts files re-keyed; qet_fr.ts filled.
lupdate check: PASS. What each of 38 languages shows, old vs new: same 118266, English instead of French 38433, comment fallback 51, differences 0.
Nine messages were never passed through tr(), so every user saw them in
French: the undo entries for grouping and ungrouping terminals, for
editing an arc (3), an ellipse, a rectangle and a PLC table in the
element editor, and the error box when integrating a title block
template fails. They are now wrapped, text unchanged.
qet_en.ts also lacked English for 63 messages added since the last
translations update (Customize dialog, mouse gestures, toolbar contents,
mirror, element scale, paste element information, paper format).
update_translations was run on qet_en.ts with lupdate's guessing
heuristics off, then the 72 new entries were translated. Most of the
diff is lupdate refreshing line numbers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit d0657f36db)
Nine messages were never passed through tr(), so every user saw them in
French: the undo entries for grouping and ungrouping terminals, for
editing an arc (3), an ellipse, a rectangle and a PLC table in the
element editor, and the error box when integrating a title block
template fails. They are now wrapped, text unchanged.
qet_en.ts also lacked English for 63 messages added since the last
translations update (Customize dialog, mouse gestures, toolbar contents,
mirror, element scale, paste element information, paper format).
update_translations was run on qet_en.ts with lupdate's guessing
heuristics off, then the 72 new entries were translated. Most of the
diff is lupdate refreshing line numbers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The Windows job configured with -DPACKAGE_TESTS=OFF, so no test ever ran
on Windows. Build the tests and run ctest after the build, with the
offscreen platform. ctest prints no output for failed tests in the MSYS2
shell, so each failed test runs again with QtTest writing its log to a
file, which is printed in its own log group. The ctest log is added to
the artifact uploaded on failure.
No change to the app or its files.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
AssignVariables::assignSequence() replaced the sequential-number
placeholders from 1 upwards with a plain QString::replace(). "%sequ_1"
is also the start of "%sequ_10", so with ten or more sequences a
formula's %sequ_10 became the first value followed by a "0". The same
held for every family (%sequf_, %seqt_, %seqtf_, %seqh_, %seqhf_,
%seqw_, %seqa_).
Run the loop from the highest number down to 1, so the longer
placeholder is gone before the shorter one is looked for.
No file-format change. Labels with fewer than ten sequences come out
exactly as before.
Tests: tst_tensequentialnumbers runs the binary's --export-bom on a
fixture whose element has the formula %sequ_10-%sequ_1 and the unit
values A to J, and expects the label "J-A". On master the label is
"A0-A".
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
Before the <sequentialNumbers> element, an element's or conductor's
sequential numbers were saved as the attributes sequ_1, sequf_1,
seqt_1, seqtf_1, seqh_1 and seqhf_1. The readers check for one of them
to take the old route, and all three lists had the same slip: sequf_1
twice, seqhf_1 never (Element::fromXml, Conductor::fromXml, and
readSequence() in the project database). A file whose only old
sequence was the hundred-folio one took the new route, found no
<sequentialNumbers>, and lost it; the database built its labels from
an empty sequence instead of refusing the fast path as it does for the
other five attributes.
Name seqhf_1 in the three lists.
No file-format change: nothing is written differently, and a file with
any of the other five attributes loads exactly as before.
Tests: tst_legacysequentialattributes runs the binary's --resave on a
fixture whose element and conductor carry only seqhf_1 (and a second
pair carrying sequ_1 as a control) and reads the <sequentialNumbers>
written back. The two seqhf_1 cases fail on master.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
BorderProperties::toXml() writes colsize and rowsize with "%1", so a
project whose default border has a width such as 60.5 saves it as
"60.5". fromXml() read both back with toInt(), which is 0 for a
decimal, and the folio clamps 0 to its 5 px minimum. New folios of such
a project got 5 px columns and rows. The folio's own reader,
BorderTitleBlock::borderFromXml(), already uses toDouble().
Read the two sizes with toDouble(&ok) and keep the previous value when
the attribute is missing, not a number, or not finite.
No file-format change: toXml() is untouched, and whole-number sizes,
which every example project has, load exactly as before.
Tests: tst_borderpropertiesxml compiles borderproperties.cpp alone and
reads 50, 60.5, 61.3 and 61.7 back, round-trips 60.5/80.25 through
toXml(), and leaves the value alone for a missing, text, nan or inf
attribute. Seven cases fail on master.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
ConductorProperties::toSettings() writes the conductor width with
QString::number(cond_size), e.g. "1.4", and fromSettings() read it back
with toInt(), which is 0 for any value that is not a whole number. A
default width of 1.4 set in the configuration became 0 on the next
start, so every new conductor was drawn with a pen of width 0.
Read it with toDouble(), and fall back to 1 when the stored value is
not a positive finite number (a hand-edited or truncated settings
file), as the other fallbacks in fromSettings() do.
No file-format change: this is the settings file only; the project
file's condsize attribute was already read with toDouble().
Tests: tst_conductorsizesetting compiles conductorproperties.cpp alone
and round-trips 2, 1.4, 0.4, 61.3 and 61.7 through toSettings() and
fromSettings() in a QSettings scope of its own; text, 0, -1, nan, inf
and a missing value give 1. Nine of the cases fail on master.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
Diagram::toXml() writes freezeNewElement and freezeNewConductor as
"true"/"false", but Diagram::fromXml() read them with toInt(), which is
0 for both words. So "freeze new elements" and "freeze new conductors"
were off again on every folio after a project was saved and reopened,
since the day the flags were added.
Compare the attribute with "true" instead. The project-level flags in
QETProject already do this.
No file-format change: the attributes are written exactly as before,
and a file without them still loads with both flags off.
Tests: tst_foliofreezeflags runs the binary's --resave on a fixture
with one folio per combination (both, elements only, none, attributes
missing) and reads the saved attributes back. The two frozen folios
fail on master.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
TitleBlockTemplateLogoManager never sets read_only_: its constructor
leaves the member uninitialised, and setReadOnly() returns early when
the new value equals the member. The template editor calls
setReadOnly() right after creating the manager, so whenever the
garbage value happened to match, the call did nothing: the add, delete
and rename buttons stayed enabled on a read-only template, and
isReadOnly() returned garbage.
Give the member a default value, false, in the header, matching the
buttons' state after construction. Only
sources/titleblock/templatelogomanager.h changes.
No behaviour change for an editable template: setReadOnly(false) on a
manager that is already editable still returns early. No file-format
change.
Tests: none of their own; an uninitialised read is what a sanitizer
finds, and this one is the same kind as the BorderTitleBlock and
ConductorProfile fixes (#1358, #1359). ctest on macOS, and all
example projects through --resave and --export-svg, unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
PartLine and PartPolygon keep their rotation in m_rot, which nothing
sets until setRotation() is called. rotation() reads it, and
setRotation() computes the difference from it, so the first rotate of
a freshly drawn line or polygon in the element editor starts from
garbage: qRound(m_rot * 100.0) on an arbitrary value is undefined
behaviour, and the angle stored afterwards is arbitrary too.
PartRectangle and PartEllipse already start at 0.
Give m_rot a default value of 0 in both headers, as PartEllipse has.
Only the two headers change.
No file-format change, and no change for a part loaded from a file or
rotated once, since both set m_rot before reading it.
Tests: none of their own; an uninitialised read is what a sanitizer
finds, and this one is the same kind as the BorderTitleBlock and
ConductorProfile fixes (#1358, #1359). ctest on macOS, and all
example projects through --resave and --export-svg, unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Beat Hangartner <beat@hangartners.ch>
Before #1310, undoing a crop left the crop rectangle in place, so a
project saved afterwards has a <crop> but shows the uncropped picture.
Since #1368 the picture shown is computed from the original and the
crop, so the first mirror or colour key of such a picture applied the
old crop again, and undoing that edit showed the cropped picture.
The picture shown is always exactly the size of the crop. fromXml() now
drops a crop that does not match it: the crop becomes the whole
original, and when the shown picture is not the original's size either,
the shown picture becomes the original. Valid files load and save as
before.
New tst_imagestalecrop: a valid crop is kept; a stale one is dropped
for a shown picture of the original's size and of another size, where a
later crop cuts the picture that was shown.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Beat Hangartner