Check and fix algorithms improvement #61531

Djedouas · 2025-04-17T09:14:53Z

Description

Following #59637 to improve the processing checks and fixes already here.

avoid using new keyword to avoid memory leaks
avoid using unnecessary pointers
remove auto keyword
better formatting
better algorithm descriptions

github-actions · 2025-04-17T10:26:29Z

🪟 Windows Qt6 builds

Download Windows Qt6 builds of this PR for testing.
(Built from commit 89eb999)

🪟 Windows builds

Download Windows builds of this PR for testing.
Debug symbols for this build are available here.
(Built from commit 89eb999)

DelazJ

Hi @Djedouas
Thanks for these updates. I was about to open issue reports for these and saw your PR, hence I'm asking here.

src/analysis/processing/qgsalgorithmcheckgeometryangle.cpp

DelazJ · 2025-04-17T11:02:02Z

src/analysis/processing/qgsalgorithmcheckgeometryangle.cpp

+  ) );
+
+  std::unique_ptr<QgsProcessingParameterNumber> tolerance = std::make_unique<QgsProcessingParameterNumber>(
+    QStringLiteral( "TOLERANCE" ), QObject::tr( "Tolerance" ), Qgis::ProcessingNumberParameterType::Integer, 8, false, 1, 13


As mentioned in qgis/QGIS-Documentation#9773, afaiu I find this tolerance parameter (8 meaning 10^-8) non-obvious, and its min/max values counter-intuitive. A parameter help is at least required for end users, but imho, we should move to something simpler as a plain double value with unit selector, as commonly used for this kind of parameter.

Yes but it's to keep consistency with Geometry Checker plugin:

output.mp4

btw, Geometry Checker accepts only an integer https://github.com/qgis/QGIS/blob/master/src/analysis/vector/geometry_checker/qgsgeometrycheckcontext.cpp#L20

Yes but it's to keep consistency with Geometry Checker plugin:

Probably, but 1E-8 as shown in the plugin looks much more meaningful than 8. I don't know how strong is/should be the link between the two tools and I'm not speaking in terms of code, rather as a Processing end user who would expect some obviousness in the options and harmonization (and that we take advantage from improved parameter types in Processing)...

Yes I agree that from an end user perspective, it could be confusing. However, we are tied with how the geometry checker is written right now.

We can't use the distance widget parameter because it would be nonsense and difficult to restrict the user to specific logarithmic values like "10, 1, 0.1, 0.001, 0.0001, etc.".

Our options are

to create another parameter type, think of its UI, and that handles the differences we have between geographic and metric distances

to change how the geometry checker tolerance is handled

I prefer to go with option 2. but in any case, it will be a work that would be done in a dedicated PR.

Right now, we could improve the documentation, or add a description of the tolerance parameter in every processing description.

I would have said

Right now, we could improve the documentation, orAND add a description of the tolerance parameter in every processing description.

They should have a start of answer in the GUI imho.

Added a short text in every processing short help.

The "Tolerance" advanced parameter defines the numerical precision of geometric operations, given as an integer n, meaning that any difference smaller than 10⁻ⁿ (in map units) is considered zero.

Given that this option is in advanced panel, I doubt people will modify it and don't think it deserves a mention in the algorithm main help. It should instead be as a tooltip over the parameter only, i.e.:

tolerance->setHelp( QObject::tr( "The \"Tolerance\" defines the numerical precision of geometric operations, " "given as an integer n, meaning that any difference smaller than 10⁻ⁿ (in map units) is considered zero." );

Djedouas · 2025-04-18T09:39:30Z

Hi @DelazJ

What do you think of changing the "Check geometry (<name>)" processing name to simply "<name>" ?

It seems heavy right now:

DelazJ · 2025-04-18T15:54:21Z

What do you think of changing the "Check geometry (<name>)" processing name to simply <name>" ?

I agree that it looks crowded. I'm not very sure that an alg named "Angle" or "Contained" would be obvious (well, it is still displayed under the check group).
Maybe "Check <name>"? though I'm still not sure if "Check contained" is a quite expressive name...

PS: I'm not sure that the alg name is supposed to be in title case... ("Check Geometry (Dangle) --> Check geometry (dangle)")
PS2: most of the time, the alg outputs are named using the alg name, i.e., Buffer --> Buffered, while here they all output "layer output" and "Error layer". It could be nice to have part of the alg name in the output(s)
PS3: Thanks for the switch to optional output.

- avoid using new keyword to avoid memory leaks - remove auto keyword - better formatting - better algorithm description

- avoid using unnecessary pointers - remove auto keyword - better formatting

QList<QgsGeometryCheckError *> contains pointers to non-Qt objects. When the list is deleted, all pointers inside the list are deleted, but every data that each pointer points to is not automatically freed.

Djedouas · 2025-04-24T08:52:56Z

Maybe "Check "? though I'm still not sure if "Check contained" is a quite expressive name...
PS: I'm not sure that the alg name is supposed to be in title case... ("Check Geometry (Dangle) --> Check geometry (dangle)")

@DelazJ In fact, I think that we need to keep the long algorithm name, because when used inside a processing model we must see a full name. I really don't have an idea about title case or not for the name, that's why there is a mix... What do you think about this?

PS2: most of the time, the alg outputs are named using the alg name, i.e., Buffer --> Buffered, while here they all output "layer output" and "Error layer". It could be nice to have part of the alg name in the output(s)

That's a good point. We could add the check name like the algorithm name:

Errors layer (check dangle)
Output layer (check dangle)

DelazJ · 2025-04-28T05:44:28Z

In fact, I think that we need to keep the long algorithm name, because when used inside a processing model we must see a full name.

@Djedouas Sorry I do not get the point with the modeler.
IMHO No need to repeat the group name in the alg, which is already displayed at the top of the alg dialog.

The name should really be shortened for the reason you mentioned earlier, and maybe revisited to greatly reflect what the alg does. Do we really need the "check..." every time? What about e.g.:

Check geometry (Angle) --> Find(/Detect) small(er?) angles
Fix geometry (Angle) --> Delete small(er?) angles

I really don't have an idea about title case or not for the name, that's why there is a mix... What do you think about this?

Actually, they should not be in title case. If you walk through the Processing toolbox you'll notice that only the first letter is in upper case, unless there is an acronym. This is the rule to follow.

That's a good point. We could add the check name like the algorithm name:

Errors layer (check dangle)
Output layer (check dangle)

I'm not really fan; still long and not inline with existing practice. What about e.g.:

Dangle errors / Errors (dangle)
Dangle / Dangle locations -- the global target of this output is not always really clear to me
Dangle fixed / Undangled (assuming it is English)

Sorry to add: There is also that great function to display algs short help when you hover over them in the toolbox. That may be helpful. But I realize there are quite a few algorithms using it. 😢

If usual Processing algs reviewers could chime in...

Djedouas added Processing Relating to QGIS Processing framework or individual Processing algorithms Geometry checker labels Apr 17, 2025

github-actions bot added this to the 3.44.0 milestone Apr 17, 2025

DelazJ reviewed Apr 17, 2025

View reviewed changes

Djedouas force-pushed the check_and_fix_algorithms_improvement branch from 45ff9b4 to 2bc04ce Compare April 18, 2025 08:35

Djedouas force-pushed the check_and_fix_algorithms_improvement branch from 2bc04ce to ed61539 Compare April 18, 2025 10:15

Djedouas added 21 commits April 24, 2025 10:30

Improve angle check processing algorithm

bb62cbb

- avoid using new keyword to avoid memory leaks - remove auto keyword - better formatting - better algorithm description

Improve area check processing algorithm

bfc15ca

- avoid using new keyword to avoid memory leaks - remove auto keyword - better formatting - better algorithm description

Improve hole check processing algorithm

6abb51f

- avoid using new keyword to avoid memory leaks - remove auto keyword - better formatting - better algorithm description

Improve missing vertex check processing algorithm

154a5f1

- avoid using new keyword to avoid memory leaks - remove auto keyword - better formatting - better algorithm description

Improve angle fix processing algorithm

394e114

- avoid using unnecessary pointers - remove auto keyword - better formatting

Improve area fix processing algorithm

d276cb4

- avoid using unnecessary pointers - remove auto keyword - better formatting

Improve hole fix processing algorithm

a1ac9e6

- avoid using unnecessary pointers - remove auto keyword - better formatting

Improve missing vertex fix processing algorithm

b7ce82b

- avoid using unnecessary pointers - remove auto keyword - better formatting

Make angle check processing output layer optional

38bd3dc

Make area check processing output layer optional

17635e2

Make contained check processing output layer optional

9ca63f2

Make dangle check processing output layer optional

4a812c9

Make degenerate polygon check processing output layer optional

13cc96d

Make duplicate check processing output layer optional

f90fb5b

Make duplicate nodes check processing output layer optional

28bc811

Make follow boundaries check processing output layer optional

0e6674c

Make gap check processing output layer optional

0fc743b

Make hole check processing output layer optional

9eb1792

Make line intersection check processing output layer optional

4e743f8

Make line layer intersection check processing output layer optional

1ad79f9

Make missing vertex check processing output layer optional

2df588a

Djedouas added 12 commits April 24, 2025 10:30

Make multipart check processing output layer optional

7fa377b

Make overlap check processing output layer optional

90c4719

Make point covered by line check processing output layer optional

2ff777c

Make point in polygon check processing output layer optional

937b383

Make segment length check processing output layer optional

3db0c91

Make self contact check processing output layer optional

2ff6357

Make self intersection check processing output layer optional

9293ebc

Make sliver polygon check processing output layer optional

8692a0b

fix(algorithmcheckgeometry): add const for geometry error loop

a222ebd

fix(algorithmfixgeometry): add verification of feedback canceled

4d2102a

fix(algorithmcheckgeometry): avoid memory leaks

b00abb1

QList<QgsGeometryCheckError *> contains pointers to non-Qt objects. When the list is deleted, all pointers inside the list are deleted, but every data that each pointer points to is not automatically freed.

fix(algorithmcheckfixgeometry): add tolerance short help text

89eb999

Djedouas force-pushed the check_and_fix_algorithms_improvement branch from 38b4cd4 to 89eb999 Compare April 24, 2025 08:33

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Check and fix algorithms improvement #61531

Check and fix algorithms improvement #61531

Djedouas commented Apr 17, 2025

github-actions bot commented Apr 17, 2025 •

edited

Loading

DelazJ left a comment

DelazJ Apr 17, 2025

lbartoletti Apr 17, 2025

lbartoletti Apr 17, 2025

DelazJ Apr 18, 2025

Djedouas Apr 18, 2025 •

edited

Loading

DelazJ Apr 18, 2025 •

edited

Loading

Djedouas Apr 24, 2025

DelazJ Apr 28, 2025

Djedouas commented Apr 18, 2025 •

edited

Loading

DelazJ commented Apr 18, 2025

Djedouas commented Apr 24, 2025

DelazJ commented Apr 28, 2025

Check and fix algorithms improvement #61531

Are you sure you want to change the base?

Check and fix algorithms improvement #61531

Conversation

Djedouas commented Apr 17, 2025

Description

github-actions bot commented Apr 17, 2025 • edited Loading

🪟 Windows Qt6 builds

🪟 Windows builds

DelazJ left a comment

Choose a reason for hiding this comment

DelazJ Apr 17, 2025

Choose a reason for hiding this comment

lbartoletti Apr 17, 2025

Choose a reason for hiding this comment

lbartoletti Apr 17, 2025

Choose a reason for hiding this comment

DelazJ Apr 18, 2025

Choose a reason for hiding this comment

Djedouas Apr 18, 2025 • edited Loading

Choose a reason for hiding this comment

DelazJ Apr 18, 2025 • edited Loading

Choose a reason for hiding this comment

Djedouas Apr 24, 2025

Choose a reason for hiding this comment

DelazJ Apr 28, 2025

Choose a reason for hiding this comment

Djedouas commented Apr 18, 2025 • edited Loading

DelazJ commented Apr 18, 2025

Djedouas commented Apr 24, 2025

DelazJ commented Apr 28, 2025

github-actions bot commented Apr 17, 2025 •

edited

Loading

Djedouas Apr 18, 2025 •

edited

Loading

DelazJ Apr 18, 2025 •

edited

Loading

Djedouas commented Apr 18, 2025 •

edited

Loading