Difference between revisions of "Documentation/Nightly/Developers/Style Guide"

From Slicer Wiki
Jump to: navigation, search
Tag: 2017 source edit
(Moved to readthedocs)
Tags: 2017 source edit, Replaced
 
Line 1: Line 1:
<noinclude>{{documentation/versioncheck}}</noinclude>
+
{{documentation/banner
= Code style =
+
| text = [https://slicer.readthedocs.io/en/latest/developer_guide/style_guide.html This page has been moved to read-the-docs.]
 
+
| background-color = 8FBC8F }}
'''Code that inherits from VTK classes should follow VTK coding conventions for naming, indentation, etc.''' See http://www.vtk.org/Wiki/VTK_Coding_Standards for details.
 
 
 
== Naming ==
 
 
 
# Acronyms should be written with the same case for each letter (all uppercase or all lowercase).
 
#: ''RASToSlice'' not ''RasToSlice''
 
#: ''vtkMRML'' not ''vtkMrml''
 
#: ''vtkSlicer'' not ''vTKSlicer''
 
# Words should be spelled out and not abreviated
 
#: ''GetWindow'' not ''GetWin''
 
# File names must follow the [http://en.wikipedia.org/wiki/CamelCase Camel case] convention
 
#: ''TestMyFeature.cxx'' not ''Test-My_Feature.cxx''
 
# Use US English words and spelling
 
#: "Millimeter" not "Millimetre"
 
#: "Color" not "Colour"
 
# Follow VTK naming conventions in VTK classes and Qt conventions in Qt classes. For example:
 
#: Use GetNumberOfXXX in VTK classes, but XXXCount in Qt classes
 
 
 
== Comments ==
 
 
 
# include extensive comments in the header files
 
# keep the comments up to date as the code changes
 
# use the keyword <code>\todo</code> to flag spots in the code that need to be revisited
 
# do not leave blocks of commented out code in the source file -- if needed insert links to prior svn versions as comments
 
 
 
== Functions ==
 
# Don't mix different levels of abstraction
 
#: Examples:
 
#:: When dealing with files names and path, use [http://vtk.org/gitweb?p=VTK.git;a=blob;f=Utilities/KWSys/vtksys/SystemTools.hxx.in;h=04f197842695c5914d1a49d4738159d3bccb08e8;hb=HEAD kwsys::SystemTools] in VTK classes, [http://doc.qt.nokia.com/{{documentation/{{documentation/version}}/qtversion}}/qfileinfo.html QFileInfo]/[http://doc.qt.nokia.com/{{documentation/{{documentation/version}}/qtversion}}/qdir.html QDir] in Qt classes or [http://docs.python.org/library/os.path.html os.path] in python. Instead of doing string manipulation manually:
 
#:::<code>QString filePath = directoryPath + "/" + fileName + ".exe"</code>)
 
#:: prefer instead:
 
#:::<code>SystemTools::JoinPath(), SystemTools::GetFilenameName()...</code>
 
#:::<code>QFileInfo(QDir directory, QString fileName), QFileInfo::suffix(), QFileInfo::absoluteFilePath()...</code>
 
#:::<code>os.path.join(), os.path.splitext(), os.path.abspath()...</code>
 
#: References:
 
#:* [http://www.amazon.com/gp/product/0132350882?ie=UTF8&tag=solisyntprog-20&linkCode=as2&camp=1789&creative=9325&creativeASIN=0132350882 Clean Code] from ''Robert C. Martin'': ''Mixing levels of abstraction within a function is always confusing. Readers may not be able to tell whether a particular expression is an essential concept or a detail. Worse, like broken windows, once details are mixed with essential concepts, more and more details tend to accrete within the functions.''
 
#:* http://zuskin.com/coding_habits__functions.htm#Abstraction_levels
 
# Use STL where you can, but follow the VTK [http://www.vtk.org/Wiki/VTK_FAQ#Can_I_use_STL_with_VTK.3F guidelines]
 
## However, prefer [http://doc.qt.nokia.com/{{documentation/{{documentation/version}}/qtversion}}/containers.html Qt Container classes] to STL classes in Qt files
 
## Note that a [http://www.vtk.org/doc/nightly/html/classvtkCollection.html vtkCollection] is somewhat equivalent to <code>std::list<vtkSmartPointer<vtkObject*> ></code>
 
 
 
== Layout ==
 
=== Includes ===
 
# Only include the necessary files, no more.
 
# Group includes per library
 
# Alphabetically sort files within groups
 
# Order groups from local to global
 
#: e.g. Module then MRML then CTK then Qt then VTK then ITK then STL
 
# Implementation files should include the header files first
 
For example:
 
  // header
 
// ...
 
// end header
 
 
#include "qSlicerMyModule.h"
 
 
// MyModule includes
 
#include "qSlicerMyModuleWidget.h"
 
#include "vtkSlicerMyModuleLogic.h"
 
 
// MRML includes
 
#include "vtkMRMLScene.h"
 
 
// Qt includes
 
#include <QDialog>
 
 
// VTK includes
 
#include <vtkSmartPointer.h>
 
 
// STD includes
 
#include <vector>
 
 
 
== Language Specific ==
 
 
 
# [[Documentation/{{documentation/version}}/Developers/Style Guide/Cpp | C++]]
 
# [[Documentation/{{documentation/version}}/Developers/Style Guide/Python | Python]]
 
# [[Documentation/{{documentation/version}}/Developers/Style Guide/CMake | CMake]]
 
 
 
== Library Dependencies ==
 
 
 
# MRML classes should only depend on vtk and itk (not Slicer Logic or Qt)
 
# Logic classes depend on MRML to store state
 
# Logic classes should encapsulate vtk and itk pipelines to accomplish specific slicer tasks (such as resampling volumes for display)
 
# GUI classes can depend on MRML and Logic and Qt
 
 
 
== Development Practices ==
 
 
 
# While developing code, enable VTK_DEBUG_LEAKS (ON by default) in your vtk build and be sure to clean up any leaks that arise from your contributions.
 
 
 
== Coordinate Systems ==
 
 
 
# World space for 3D Views is in RAS (Right Anterior Superior) space. See [[Coordinate systems]].
 
# All units are expressed in Millimeters (mm)
 
 
 
== String encoding: UTF-8 everywhere ==
 
 
 
Slicer follows uses [https://utf8everywhere.org/ UTF-8 everywhere]: all strings in std::string, char[] arrays, files, etc. are in UTF-8 (except in rare exceptions where this is very clearly indicated). We don't use code pages or any other unicode encoding. If this leads to incorrect behavior anywhere then the underlying issue must be fixed (e.g., if a VTK function does not work correctly with UTF-8 encoded string input then a fix has to be submitted to VTK).
 
 
 
On Windows, process code page of all Slicer executables (main application, CLI modules, tests, etc.) are explicitly set to UTF-8 by using "ctk_add_executable_utf8" function instead of plain "add_executable" in CMake. This makes all standard API functions to use UTF-8 encoding, even in third-party libraries. This mechanism requires Windows Version 1903 (May 2019 Update) or later. On Linux and Mac, encoding is already expected to be UTF-8 (it is currently not checked or enforced in any way).
 
 
 
Conversion from std::string to QString:
 
  std::string ss = ...
 
  QString qs1 = QString::fromUtf8(ss); // this is slightly preferred, as it is very clear and explicit
 
  QString qs2 = QString(ss); // same result as fromUtf8, acceptable, as it is a bit simpler and used throughout the code base anyway
 
 
 
Conversion from QString to std::string:
 
  QString qs = ...
 
  std::string ss  = QString::toUtf8(qs);
 
 
 
Printing to console: in general, VTK, Qt, or ITK logging macros are preferred but if for some reason text must be printed on console then use qPrintable macro. This macro converts the string
 
  std::cerr << "Failed to create file " << qPrintable(filePath) << std::endl;
 
 
 
Qt logging macros (qDebug, qWarning, QFatal): these macros expect UTF-8 encoded strings, therefore do not use qPrintable macro
 
 
 
File management: All filenames have to be passed to file functions (such as fopen) must be UTF-8 encoded. All text file content is expected to be UTF-8 encoded, except very rare cases when a different encoding is explicitly specified in the file (for example in incoming DICOM files may use different encoding).
 
 
 
== Error and warning messages ==
 
{| width="100%"
 
| valign="top"|
 
The ITK, VTK, Qt, std::cout, std::cerr .. all appear in the error log and can easily be filtered according to their type (debug/warning/error).
 
 
 
* '''Errors''': Error should be used to signal something that should not happen. They usually mean that the execution of the current function/code should be stopped.
 
* '''Warnings''': Warning should be used to signal potentially dangerous behavior. Also consider using these in deprecated methods to warn your fellow developers.
 
* '''Debugs''': For general debug and developer aimed information, one can use the debug messages.
 
 
 
| align="right"|
 
|[[Image: Slicer4ErrorLog.jpg|thumb|200px| Error log in Slicer 4]]
 
|}
 
 
 
* In Qt-based classes:
 
** For error messages, use [http://qt-project.org/doc/qt-4.8/qtglobal.html#qCritical qCritical()].
 
  if (somethingWrongHappened)
 
    {
 
    qCritical() << "I encountered an error";
 
    return;
 
    }
 
 
 
:* For warnings, use [http://qt-project.org/doc/qt-4.8/qtglobal.html#qWarning qWarning()].
 
  qWarning() << "Be careful here, this is dangerous";
 
 
 
:* For debug, use [http://qt-project.org/doc/qt-4.8/qtglobal.html#qDebug qDebug()]:
 
  qDebug() << "This variable has the value: "<< value;
 
 
 
* In VTK-based classes:
 
** For error messages, use [http://www.vtk.org/doc/release/3/html/vtkSetGet_8h.html vtkErrorMacro()].
 
  if (somethingWrongHappened)
 
    {
 
    vtkErrorMacro("I encountered an error");
 
    return;
 
    }
 
 
 
:* For warnings, use [http://www.vtk.org/doc/release/3/html/vtkSetGet_8h.html vtkWarningMacro()].
 
  vtkWarningMacro("Be careful here, this is dangerous");
 
 
 
:* For debug, use [http://www.vtk.org/doc/release/3/html/vtkSetGet_8h.html vtkDebugMacro()]:
 
  vtkDebugMacro("This variable has the value: "<< value);
 
 
 
== Misc. ==
 
# Ideally, no more than 80 characters per line.  Keep line length under 120 characters.
 
 
 
= Commits =
 
==Summary==
 
 
 
* Separate the subject from body with a blank line
 
* Limit the subject line to 50 characters
 
* Capitalize the subject line
 
* Do not end the subject line with a period
 
* Use the imperative mood in the subject line
 
* Wrap the body at 72 characters
 
* Use the [[#Message_content|body to explain]] what and why vs. how
 
** If there was important/useful/essential conversation or information, copy or include a reference
 
* [[#Commit_message_prefix|Prefix]] the commit message title with "BUG:", "COMP:", "DOC:", "ENH:", "STYLE:". Note the ':' (colon) character.
 
* When possible, one keyword to scope the change in the subject (i.e. "STYLE: README: ...", "BUG: vtkMRMLSliceLogic: ...")
 
 
 
==Commit message prefix==
 
Subversion Commits to Slicer require commit type in the comment.
 
 
 
Valid commit types are:
 
  BUG:  - a change made to fix a runtime issue
 
            (crash, segmentation fault, exception, or incorrect result,
 
  COMP:  - a fix for a compilation issue, error or warning,
 
  DOC:  - a documentation change,
 
  ENH:  - new functionality added to the project,
 
  PERF:  - a performance improvement,
 
  STYLE: - a change that does not impact the logic or execution of the code.
 
            (improve coding style, comments).
 
Note that the ':'(colon) directly follows the commit tag. For example, it is: "STYLE:" not "STYLE :"
 
 
 
The Subversion command to commit the change is:
 
  svn commit -m "BUG: fixed core dump when passed float data" filename1[, filename2, ...]
 
 
 
By using the <code>-m</code> command line option, it's not possible to submit a message having multiple line.
 
Submitting a mutli-line message can be achieved using the <code>-f</code> option:
 
  svn commit -f /path/to/message filename1[, filename2, ...]
 
 
 
It's also possible to set the environment variable [http://www.google.com/search?q=SVN_EDITOR|<code>SVN_EDITOR</code>]
 
 
 
==Message content==
 
# A good commit message title (first line) should '''explain what the commit does for the user, not ''how'' it is done'''. ''How'' can be explained in the body of the commit message (if looking at the code of the commit is not self explanatory enough).
 
#: Examples:
 
#:* Bad: <code>BUG: Check pointer validity before dereferencing</code> -> ''implementation detail'', ''self-explanatory'' (by looking at the code)
 
#:* Good: <code>BUG: Fix crash in Module X when clicking Apply button</code>
 
#:* Bad: <code>ENH: More work in qSlicerXModuleWidget</code> -> <code>more work</code> is ''too vague'', <code>qSlicerXModuleWidget</code> is too ''low level''
 
#:* Good: <code>ENH: Add float image outputs in module X</code>
 
#:* Bad: <code>COMP: Typo in cmake variable</code> -> ''implementation detail'', ''self-explanatory''
 
#:* Good: <code>COMP: Fix compilation error with Numpy on Visual Studio</code>
 
# If the commit is related to a [http://na-mic.org/Mantis/view_all_bug_page.php mantis issue] (bug or feature request), you can mention it in the commit message body by preceding the issue number with a '''#'''(pound) character:
 
BUG: Fix crash in Volume Rendering module when switching view layout
 
 
vtkSetAndObserveMRMLNodeEventsMacro can't be used for observing all types of vtkObjects,
 
only vtkMRMLNode is expected by vtkMRMLAbstractLogic::OnMRMLNodeModified(...)
 
Closes #1641
 
Where <code>1641</code> refers to the [http://www.na-mic.org/Bug/view.php?id=1641 issue number] in mantis.
 
 
 
# Notice the empty 2nd line.
 
 
 
==Importing changes from external project/repository==
 
When you update the git tag or svn revision of any external project, explain in the commit message what the update does instead of just mentioning that an update in made.
 
 
 
This will avoid having a Slicer commit history made of unintelligible messages:
 
r19180 - ENH: Update git tag
 
r19181 - BUG: Update svn revision
 
r19182 - ENH: revision updated
 
...
 
 
 
Ideally it should be the same message than the commit(s) in the external repository.
 
 
 
Read [[Documentation/{{documentation/version}}/Developers/Versioning#Project_fork]] for an exhaustive list of recommendations.
 
 
 
Example:
 
 
 
{{pre2|<nowiki>
 
COMP: Update MultiVolumeExplorer to fix unused-local-typedefs warnings
 
 
 
$ git shortlog 17a9095..d68663f --no-merges
 
Jean-Christophe Fillion-Robin (1):
 
      COMP: Fix unused-local-typedefs warnings
 
</nowiki>}}
 
 
 
See [http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=23377 r23377]
 
 
 
== Resources ==
 
 
 
* Read more on [http://chris.beams.io/posts/git-commit/ How to Write a Git Commit Message]
 
* Discussion section of [http://git-scm.com/docs/git-commit git-commit(1)]
 
 
 
<!-- http://www.na-mic.org/Wiki/index.php/Engineering:Subversion_Repository -->
 
 
 
= UI Design Guidelines =
 
{{:Documentation/{{documentation/version}}/Developers/Style Guide/UI}}
 
 
 
= Logging =
 
 
 
The following log levels are used in Slicer:
 
* Error: detected errors, conditions that should never occur
 
* Warning: potential errors, possible computation inaccuracies
 
* Info: important events, application state changes (helps to determine what the steps lead to a certain error or warning); one user action should not generate more than 1-2 info level messages
 
* Debug: any information that may be useful for debugging and troubleshooting
 
 
 
In VTK classes:
 
* vtkErrorMacro("vtkMRMLClipModelsNode:: Invalid Clip Type");
 
* vtkWarningMacro("Model " << modelNode->GetName() << "'s display node is null\n");
 
* vtkDebugMacro("CreateWidget: found a glyph type already defined for this node: " << iter->second);
 
 
 
In QT classes:
 
* qCritical() << "qSlicerUtils::setPermissionsRecursively: Failed to set permissions on file" << info.filePath();
 
* qWarning() << "qSlicerIOManager::openScreenshotDialog: Unable to get Annotations module (annotations), using the CTK screen shot dialog.";
 
* qDebug() << "qMRMLSceneFactoryWidget::deleteNode(" <<className <<") no node";
 
 
 
In Python:
 
* logging.error("This is an error message. It is printed on the console (to standard error) and to the application log.")
 
* logging.warning("This is a warning message. It is printed on the console (to standard error) and to the application log.")
 
* logging.info("This is an information message. It is printed on the console (to standard output) and to the application log.")
 
* logging.debug("This is a debug message. It is only recorded in the application log but not displayed in the console. File name and line number is added to the log record.")
 
 
 
= Wiki =
 
 
 
* Prefer to follow [http://en.wikipedia.org/wiki/Wikipedia:Naming_conventions_(capitalization) Wikipedia conventions] for page naming: ''For multiword page titles, leave the second and subsequent words in lowercase unless the title phrase is a proper noun that would always occur capitalized, even in the middle of a sentence''.
 
* Versioned pages should include '<nowiki><noinclude>{{documentation/versioncheck}}</noinclude></nowiki>' at the very beginning.
 
** FAQ pages should also have (immediately after the above): '<nowiki><noinclude>__TOC__={{#titleparts: {{PAGENAME}} | | -1 }}=</noinclude><includeonly>='''</nowiki>''name of parent FAQ''<nowiki>: {{{1}}}'''=</includeonly></nowiki> (be sure to replace "''name of parent FAQ''").
 

Latest revision as of 20:32, 15 March 2022

Home < Documentation < Nightly < Developers < Style Guide