All Projects

QCAD Professional
Win7 winInstaller 32bit
Version: 3.25.2.7 (3.25.2.7)
Internet: QCAD.org
Build Date: Dec 16 2020
Revision: e06f230
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

Include file was open when using ‘Edit Block In-place (BB)’ in an other drawing.
Happend on 5/1/2021 just before 2:56 am
https://qcad.org/rsforum/viewtopic.php?&t=7945#p31055

‘Edit Block In-place (BB)’ is never used in this drawing.
There was never a Block in this file.
I might have switched focus.

Re-installed QCAD a dozen times meanwhile.
Wiped all what remained of QCAD twice.
Cleaned up the windows registry twice.
Lastly left out all kinds of my custom scripts.
Simply a bare install with the msi file.

How to replicate:
- Start QCAD in normal mode
- Load included file
- Set App.Prefs .. Startup .. Restore open files
- Close QCAD
- Start QCAD in debugger mode
(-rescan -enable-script-debugger -always-load-scripts)

Boot fails.
Every successive trial.

When booting in normal mode I see the BB options toolbar blink shortly.

3 Screengrabs in DebuggerModeError.zip in the same topic.
The blink I can’t catch.

I can Boot with no file to retore on open in debugger mode and load the file.
I have several of those files, all were open that day at 2:56 am

CVH

(Re-)combining polys touching at node with ‘create from selected’ leads to application crash.

QCAD 3.20.1 Win7 32bit

In the file example 2.
Regards

Trimming polylines especialy those with a string of small segments should be done at the intersection of those polylines.
Now trimming is done with the seggments that where the nearest when selecting the two polylines.

Scaling without keeping the aspect ratio fails to scale an arc to an ellips when this is part of a polyline.

Discovered while documenting feature request ‘Skewing’

CVH

Andrew,

Usage states:
“Default is to explode everything into points, lines, arcs and circles”

In this Forum Topic 10358 you explain to use an exclusion list:
“dwgexplode is geared towards exploding everything. You’d have to provide a list of all entity types that are acceptable (i.e. don’t need to be exploded)”

Your example list includes the entity type ‘Splines’.
Marking splines entities as acceptable or not contradicts that they are never exploded.

But in the code itself is stated:
“explode texts to splines, arcs, lines”

TTF text is indeed exploded in mostly splines and some arcs or lines if that is appropriate.

All depends on the intended behavior ...
Related forum topic:
https://www.qcad.org/rsforum/viewtopic.php?f=32&t=10767

Regards,
CVH

Andrew,

It seems that Attribute Definitions are created by default with Invisible = Yes
I can’t control this property on creation.
I can select the newly created Attribute Definition and toggle the flag.

Consequence when not updating this flag for each new Attribute Definition is that inserted Block Attributes disappear on synchronizing.
See attached files.

The default Invisible = Yes setting is not retained on save/reload.
(You will probably discover these flags to be ‘No’ in the attached files)
Synchronizing after a reload has no longer the mentioned consequence.

On inserting a Block Reference based on a Block with Attribute Definitions the Invisible flag of new Block Attributes is always ‘No’.
InsertBlock.js nor InsertBlockItem.js consider the Attribute Definition Invisible flag status when creating and inserting a new RAttributeEntity object.
e.g.: InsertBlock.js L184-L206

IMHO:
- Invisible = Yes on creation is not the best default for new Attribute Definitions.
- Not retaining this flag on save/reload is probably not intended.
- Disregarding the Attribute Definitions Invisible flag on block insertion seems incorrect.

Related forum topic:
https://qcad.org/rsforum/viewtopic.php?f=32&t=10714

Regards,
CVH

Andrew,

Related topic:
https://www.qcad.org/rsforum/viewtopic.php?f=8&t=10562

From the looks of it the cursor image data is read at a wider pace as it is stored.
One can distinguish the double cross positioning QCAD cursor but it is shifted.

Regards,
CVH

Andrew,

Related topics:
https://www.qcad.org/rsforum/viewtopic.php?f=33&t=10567 https://www.qcad.org/rsforum/viewtopic.php?f=32&t=10563

Regards,
CVH

Andrew,

In some cases Draw > Ellipse > Inscribed in Quadrilateral (EQ) may return a perfectly healthy result but rotated for 90 degrees.
- Size doesn’t matter.
- Quadrant doesn’t matter.
- Geometrical open or closed base shape doesn’t matter.
- A mirrored base shape usually preforms well.

See attachment with examples.
The incorrect results (red) are rotated 90 degrees around the ellipse center (dotted)
Green is a correct result from a mirrored base shape.

Regards,
CVH

Andrew,

Discovered in writing a custom tool:
hatchEntity.setCustomProperty(”MijnEPB”, “Ruimte”, roomData.numberText);
Works perfectly and I can retrieve the room numbers of all semi-automatically placed hatches in this instance of QCAD.

But I can not store and reload this file with these custom properties to use them in a new QCAD session.
I then assume the title must be ‘QCAD’.
Then I wonder how QCAD handles custom properties by other CAD applications.
Are they also removed on loading/writing?

QCAD doesn’t list these custom properties in the Property Editor.
Here hidden was fine unless they are not stored with the document.

Is it even possible to store them in the drawing file?
Why is the title a required variable when it can’t differ from ‘QCAD’?
Would it be possible to list them up in the Property Editor because I know they are filtered on title ‘QCAD’.

For now I revert to:
hatchEntity.setCustomProperty(”QCAD”, “EPB-Ruimte”, roomData.numberText);
Similar as the CAM tool that uses ‘Cam....’ for all keys.

Regards,
CVH

Andrew,

Deleting a node of a polyline (OD) followed by Undo (OO) while OD is still in ChoosingNode state and then deleting another node returns a polyline where both nodes are removed.

this.polylineEntity is set once in pickEntity() altered after pickCoordinate() and a clone is returned to the drawing.
The next pickCoordinate() continues with altering this.polylineEntity without retrieving the current form and is thus unaware that the former action might be undone.

The work around to undo a removal is first to terminate (QQ) PolylineDeleteNode then Undo (OO) and then start over with selecting the polyline and indicating nodes.

While adding nodes (OA) we can simply undo the last action and continue adding nodes.

Because the new entity has the same id as the initially one ...
... It should be possible to re-retrieve the actual entity by id in pickCoordinate() or deleteNode().

Regards,
CVH

Andrew,

With 3 co-linear points:
- Arc3P will warn the user that an arc can’t be created (Line 141).
- Circle3P creates a null-circle without issuing a similar warning (Line 149).

The difference sits in ShapeAlgorithms.js
- ShapeAlgorithms.createArcFrom3Points Line 1465 checks if the returned arc isValid.
- ShapeAlgorithms.createCircleFrom3Points Line 1436 checks if the returned circle isNull.

Similar as RArc.cpp at Line 108 RCircle.cpp returns an RCircle() at Line 74 when there is no intersection of the two middle lines.
An RCircle() isn’t a valid circle as such but it is not nothing.

The error from ShapeAlgorithms.js is also treated somewhat different in the pickCoordinate event of Arc3P vs Circle3P.

Regards,
CVH

Andrew,

When we explode a non-uniform scaled block with arcs, the returned ellipses do not inherent the general properties of the block like other entities do.
They are exploded use the default properties that we can alter in the PEN toolbar for example.

Regards,
CVH

Andrew,

Hatches in uniform scaled blocks explode perfect among other entities, even a hatch pattern is scaled.
Hatches in non-uniform scaled blocks are exploded as not scaled.

One could scale the hatch boundaries non-uniform.
For solid hatches the pattern scale doesn’t matter.
For patterned hatches the pattern scale and angle is more problematic.
But why not explode it as is, the pattern is transformed as expected for the non-uniform scaled block reference.

Or one doesn’t transform the pattern at all but then the the non-uniform scaled block reference isn’t displayed correctly.

Regards,
CVH

Andrew,

Seems to be a copy/paste but never re-edited kind of mistake.

Quick fix:
https://github.com/qcad/qcad/blob/master/scripts/Misc/MiscInformation/InfoCentroids/InfoAreaCentroid/InfoAreaCentroid.js#L1320 line 1320 should read:
return [centerPoint.x, centerPoint.y, centerPoint, radius*radius * Math.PI];

line 1318 was intended to replace the ‘area’ portion of line 1320 literally.
As is, the code in this line makes no sense and line 1318 should be removed.

Still have to investigate if the area of a full 360° arc sector/segment should be signed ...
... and if RS.AngleTolerance is required when comparing.

Regards,
CVH

Andrew,

simple_modify.js
function paste()
10 * \param docDestination RDocument to paste into (or undefined for current document)

A) This should be a document interface of the type RDocumentInterface.
B) The function uses the parameter diDestination and not docDestination

https://www.qcad.org/rsforum/viewtopic.php?f=30&t=9742

Regards,
CVH

Additional to  FS#2425 

Include scale option in Option Toolbar for Arc Length Dimensions.
Include scaling RDimArcLengthData by linear factor.

Regards,
CVH

Andrew,

In “Loop not closed at point 120.079/259.997.” we can not use the coordinate as such.
With my preferences I would need 120,079;259,997 as a valid coordinate to enter in the command line.
I do this ‘translation’ in between in Notepad.

Not that personally my boundaries are frequently off

Reported by <OttoZ> in:
https://www.qcad.org/rsforum/viewtopic.php?f=89&p=38836#p38829
‘Die Koordinaten lassen sich nicht kopieren, sonst könnte man zur Fehlstelle springen.’

In all cases the “/” is not a valid Cartesian coordinate separator.
The dot at the end of the sentence is also not so common in QCAD.

Regards,
CVH

Andrew,

When selecting a Hatch entity:
For a solid filled Hatch the boundaries total length is given.
With a pattern that returns the summed length of the patterned lines.

I think that the recent performance update for RHatchData::getLength returns something different than intended.

Regards,
CVH

Andrew,

Trace Bitmap requires a selection.
QCAD stalls when the first entity id is not an RImageEntity .

The prevImageEntity.getFileName() in the beginEvent will throw an exception for all but an RImageEntity.

The odds are against us that in a multiple selection an image is the first id.

I think you need to filter on image entities like later in the applyOperation function.

Regards,
CVH

Seen
https://github.com/qcad/qcad/commit/302010556a077f71b3de6c3c0f8f3df0282a5da9 And that all entries are a division by 1000

It is exactly 9 460 730 472 580 800 meter
To return a correct factor to meters the value must be a factor 1000 larger.

131 return 9460730472580800000 / 1000;

More info:
https://www.qcad.org/rsforum/viewtopic.php?f=31&t=9506&p=38295#p38295

Kind regards,
CVH

Tested on a Win7 32bit and a Win10 64bit system.

With App.Prefs defaults for new drawings set to mm, mm, metric.
Inserting a ‘CubeCuttingOut’ with the default size 10 into a drawing in inch results in size 0.39370079” what is indeed 10mm.

One would expect it to be sized 10 units of the target drawing.

Recently reported by Husky concerning the QCAD Utilities by MikeJ.
https://qcad.org/rsforum/viewtopic.php?f=30&t=9429&start=15#p38045

ScriptItems are first created on a new (offline) document.
QCAD will then adapt the size with unit conversion.
Then it will scale, rotate, flip according the ‘Insert Script’ tool options.

This is fully predictable for standard library items where the source is a dxf file with a defined unit and no unit conversion would occur if the library source unit was ‘None’.

I then assume that this can be fixed with setting the new RDocument in InsertScriptItem.prototype.beginEvent its unit to ‘None’.
Not knowing what the intended behavior should be for any other case, source or usage ...

Regards,
CVH

QCAD Professional Win 7 WinInstaller

Version: 3.27.0.0 (3.27.0)
Internet: QCAD.org
Build Date: Nov 30 2021
Revision: 62c65de
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

Related topic:
https://qcad.org/rsforum/viewtopic.php?f=31&p=37972#p37964

In script SaveAs.js on line 104 the sourced filename is correct.
When this is not empty on line 147 the fileDialog should select this file in its own path.
The testing sequence in the topic shows this is not always the case.
I am using the system file dialog.

Firts setting the corresponding folder for the fileDialog seems to fix this:
147 if (fileName.length!==0) {
148 fileDialog.setDirectory(fileInfo.absolutePath());
149 fileDialog.selectFile(fileInfo.absoluteFilePath());
150 }
151 else {
....

Regards,
CVH

Andrew,
presumed not to be OS/version depending.

Referring to topic: https://qcad.org/rsforum/viewtopic.php?f=33&t=9073.
An issue that has similarities with several topics,
and some bug reports (closed or still open).

• https://qcad.org/rsforum/viewtopic.php?t=8406
• https://qcad.org/rsforum/viewtopic.php?t=8343
• https://qcad.org/bugtracker/index.php?do=details&task_id=2063

Solid hatches in blocks that are scaled up seems to be crippled.
Very pronounced with arcs in the boundary.
I suspect that the arc interpolation isn’t scaled along with the Block Reference scale.

Another issue exists for patterned hatches in blocks.
I suspect that the hatch origin is kept as defined in the block. Applying the block reference insertion point and at scale as an offset to the hatch origin while rendering would:
- hatch the inserted block exactly the same as it is created in the block.
- reduce lagging and increase pattern accuracy/coherency there the hatch origin is at the same distance to the hatch as it is in the block.
I know this doesn’t solves anything when the block is created from entities and the hatch origin wasn’t close to the hatch itself.

Regards,
CVH

Andrew,

Encountered with snapshot 3.26.4.9
On both a 32bit win7 and a Linux machine.
But probably not related to version nor OS.

The calculated area centroid position for a self-intersecting shape is invalid.
Therefor polylines and splines are checked for their self intersection nature.
Polylines with line segments require a tolerance of 1e-6.
For polylines with arc segments there is no easy to formulate value.
Reverted to something less strict than the arc connect tolerance: 1.1e-3
Refer to: https://www.qcad.org/rsforum/viewtopic.php?f=30&t=8526&p=34056#p33694

RSpline.getSelfIntersectionPoints() has no tolerance value.

Some perfect valid spline shapes fail this test.
Not related to the XP spline approx tolerance preference.
Related to orientation, try the failing shape when reversed in situ.
Related to their position in the plane, see translated examples.

Reproducible with control point splines.
Reproducible when rotating the shape.
Such failing closed splines can be hatched and an area centroid is returned.

Regards,
CVH

PS: Got a notice posting this:
“Notice: Trying to access array offset on value of type bool in /home/qcadch/public_html/bugtracker/scripts/details.php on line 222”

Windows 7 pro 32 bit msi-Installer
Version: 3.26.4.5 (3.26.4.5)
Internet: QCAD.org
Build Date: Jul 20 2021
Revision: e596276
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

Andrew,

With an CW ellipse Arc > 180° the GUI method info Arc/Circle/Ellipse (IC) in segment mode throws an exception:
Uncaught exception at :scripts/Information/InfoArcCircleArea/InfoArcCircleArea.js:210: ReferenceError: Can’t find variable: remainSectorArea

Rather a complex way to get the area of an ellipse arc segment.

There: area = majorR * minorR / 2 * (sweep - Math.sin(sweep)).
- where majorR, minorR & sweep can simply be sourced from the ellipse arc shape/entity.
- where the area is signed ... +CCW, -CW

About identical as for an arc segment.

For an (ellipse) arc sector one would only need to add/subtract
the area of triangle |StartP-CenterP-EndP|

Regards,
CVH

QCAD Professional
Win7pro Win-Installer
Version: 3.26.4.5 (3.26.4.5)
Internet: QCAD.org
Build Date: Jul 20 2021
Revision: e596276
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

Trimming With AX slows down with many text entities.
Seeing the same lagging with D2 on Circle entities.

Also seen on Windows 10 Pro 3.26.4.0
Forum topic: https://qcad.org/rsforum/viewtopic.php?t=8729

Attached files:
TrimAuto_1.dxf ... 11 entities ... trimming instantaneous
TrimAuto_2.dxf ... 11 entities + 1936 LINES ... rather instantaneous
TrimAuto_3.dxf ... 11 entities + 1936 ARCS ... less than 1 second
TrimAuto_4.dxf ... 11 entities + 121 TEXT ... over 3 or 4 seconds
Slower for larger Arc / Circle

The questions here are:
Should AX trim at text entities?
Can this be optional?

Regards,
CVH

QCAD Professional
Win7Pro WinInstaller
Version: 3.26.4.5 (3.26.4.5)
Internet: QCAD.org
Build Date: Jul 20 2021
Revision: e596276
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

See attached file:
- The 9 center snapping points of the 24 bulging segments are all presumed to be incorrect
- For some segments the center snap is outside the arc shape from a bulging segment
- For some segments there are 2 solutions

Regards,
CVH

Andrew,

Seen in al previous versions that I know of.
To recreate and merely as an example:
With:
- A polyline entity on a layer ‘polyline’ (notLocked - notFrozen - Visible)
- A hidden layer ‘exploded’ (notLocked - notFrozen - notVisible)

- Select the polyline .. DP .. XP » duplicated polyline explosion in selection.
- Set the selection its layer to ‘exploded’ by a double click in the Layer List or by Property Editor.
- Hide layer ‘polyline’ by toggling its visibility.
- Unhide layer ‘exploded’ by toggling its visibility.

The explosion is still in a dark red color.
Workaround: TA & TN

Regards,
CVH

Andrew,

With an RPolyline shape.
That is trimmed by RPolyline.trimEndPoint(cutPoint1, RVector.invalid, false);
And what is turned into a null-shape having only one single vertex.
Getting the end orientation of such a shape lead to an Application Crach.

RPolyline.getDirection1(); Doesn’t fail and returns 0.0rad.
RPolyline.getDirection2(); Fails critical.

I would have expected 0.0rad or ‘undefined’ for both.

One can circumvent that with simply testing the nature of the RPolyline before getting the end direction, and this each time required.

Regards,
CVH

Andrew,

Besides that Misc .. Information .. ‘Store Positions’ - stores the startpoint of each polyline
- or stores all nodes of each polyline (With: InfoStorePositions/PolylineNodes = true)
Refering to https://qcad.org/rsforum/viewtopic.php?t=6300

I somethimes require the midpoints of the line and arc segments as spline controlpoints
+ the start and the endpoint when geometrical open.

Could that be made an undocumented preference too?

Regards,
CVH

QCAD Professional
Win7 32bit wininstaller
Version: 3.25.2.13 (3.25.2.13)
Internet: QCAD.org
Build Date: Feb 12 2021
Revision: cad2d36
Qt Version: 5.13.2
Architecture: i386
Compiler: Unknown

Andrew,

Break out Manual (B2) on Arc and open Polyline results
seems to depend on the breakpoint picking order.

A segment is broken out when the picking order is in the entity orientation.
The intended behavior.

That differs when the picking order is opposite the entity orientation.
One would expect that the picking order matters when removing
a segment between the picked points.

Example file included

Noticed while merging code to PolyLineBreakSymbol.js
https://github.com/qcad/qcad/blob/master/scripts/Misc/MiscDraw/PolyLineBreakSymbol/PolyLineBreakSymbol.js Line 300-312 orders cut points on distance from start.
Otherwise, the results where inconsistent.

Regards,
CVH

Andrew,
This is bugging me for over a year now.
I filed it as a feature request as it is dismissed twice as a possible bug on the forum.
In both cases the advice is to only use Specific Scale 0 for Leaders.
(... and adapt all other dimensions to that)

I am aware that an arrow head has to fit twice on the first leg of a Leader Entity.

The size is set global in the Drawing Preferences under Dimension Settings: ‘Text Height’.
Under Dimension Settings, there is also a global ‘Size scale factor’.
And every dimension entity has a Specific Property called ‘Scale’ that supersedes this global ‘Size scale factor’.

Proof of concept: Dim size 5 and for any regular dimension.
1) Size scale factor 1, Specific Scale 0 » text height & arrow length = 5 units.
2) Size scale factor 2, Specific Scale 0 » text height & arrow length = 10 units.
3) Size scale factor 2, Specific Scale 1 » text height & arrow length = 5 units.
4) Size scale factor 1, Specific Scale 2 » text height & arrow length = 10 units.
5) Size scale factor 10, Specific Scale 0.5 » text height & arrow length = 2.5 units.
6) Size scale factor 99, Specific Scale 0.1 » text height & arrow length = 0.5 units.
→ Obvious is that the ‘Specific Scale’ supersedes the dimensions ‘Size scale factor’.

Those two last Leader arrows (2.5 & 0.5 long) would definitely fit twice on an 8 units long first Leader leg! And they do perfectly while drawing them.

However, the arrow will be reset when the file is saved & loaded.

For case 5 the first leg has to be (2 times 5×10) 100 units or over so an arrow 2.5 long remains on save (=40 times). In the absurd case 6 that would be (2 times 5×99) 990 units or over for an arrow only 0.5 units long (=1980 times).
Verfied and true.

It seems that on save/load:
The Specific Scale is not taken into consideration while evaluating ‘At least twice’.

Apart from that they render fine when first drew.
And reappear when setting the arrow back on again after load.

Regards,
CVH