PIKApp/devel-docs/README.md

24 KiB

title
Developers documentation

This manual holds information that you will find useful if you develop a PIKA plug-in or want to contribute to the PIKA core.

People only interested into plug-ins can probably read just the Plug-in development section. If you wish to contribute to all parts of PIKA, the whole documentation is of interest.

[TOC]

Plug-in development

Concepts

Basics

Plug-ins in PIKA are executables which PIKA can call upon certain conditions. Since they are separate executables, it means that they are run as their own process, making the plug-in infrastructure very robust. No plug-in should ever crash PIKA, even with the worst bugs. If such thing happens, you can consider this a core bug.

On the other hand, a plug-in can mess your opened files, so a badly developed plug-in could still leave your opened images in an undesirable state. If this happens, you'd be advised to close and reopen the file (provided you saved recently).

Another downside of plug-ins is that PIKA currently doesn't have any sandboxing ability. Since we explained that plug-ins are run by PIKA as independant processes, it also means they have the same rights as your PIKA process. Therefore be careful that you trust the source of your plug-ins. You should never run shady plug-ins from untrusted sources.

PIKA comes itself with a lot of plug-ins. Actually nearly all file format support is implemented as a plug-in (XCF support being the exception: the only format implemented as core code). This makes it a very good base to study plug-in development.

Procedural DataBase (PDB)

Obviously since plug-ins are separate processes, they need a way to communicate with PIKA. This is the Procedural Database role, also known as PDB.

The PDB is our protocol allowing plug-ins to request or send information from or to the main PIKA process.

Not only this, but every plug-in has the ability to register one or several procedures itself, which means that any plug-in can call features brought by other plug-ins through the PDB.

libpika and libpikaui

The PIKA project provides plug-in developers with the libpika library. This is the main library which any plug-in needs. All the core PDB procedures have a wrapper in libpika so you actually nearly never need to call PDB procedures explicitly (exception being when you call procedures registered by other plug-ins; these won't have a wrapper).

The libpikaui library is an optional one which provides various graphical interface utility functions, based on the PIKA toolkit (GTK). Of course, it means that linking to this library is not mandatory (unlike libpika). Some cases where you would not do this are: because you don't need any graphical interface (e.g. a plug-in doing something directly without dialog, or even a plug-in meant to be run on non-GUI servers); because you want to use pure GTK directly without going through libpikaui facility; because you want to make your GUI with another toolkit…

The whole C reference documentation for both these libraries can be generated in the main PIKA build with the --enable-gi-docgen autotools option or the -Dgi-docgen=enabled meson option (you need to have the gi-docgen tools installed).

TODO: add online links when it is up for the new APIs.

Programming Languages

While C is our main language, and the one libpika and libpikaui are provided in, these 2 libraries are also introspected thanks to the GObject-Introspection (GI) project. It means you can in fact create plug-ins with absolutely any language with a GI binding though of course it may not always be as easy as the theory goes.

The PIKA project explicitly tests the following languages and even provides a test plug-in as a case study:

  • C (not a binding)
  • Python 3 (binding)
  • Lua (binding)
  • Vala (binding)
  • Javascript (binding, not supported on Windows for the time being)

One of the big advantage of these automatic bindings is that they are full-featured since they don't require manual tweaking. Therefore any function in the C library should have an equivalent in any of the bindings.

TODO: binding reference documentation.

Note: several GObject-Introspection's Scheme bindings exist though we haven't tested them. Nevertheless, PIKA also provides historically the "script-fu" interface, based on an integrated Scheme implementation. It is different from the other bindings (even from any GI Scheme binding) and doesn't use libpika. Please see the Script-fu development section.

Tutorials

TODO: at least in C and in one of the officially supported binding (ideally even in all of them).

Porting from PIKA 2 plug-ins

Debugging

PIKA provides an infrastructure to help debugging plug-ins.

You are invited to read the dedicated documentation.

Script-fu development

Script-fu is its own thing as it is a way to run Scheme script with PIKA. It is itself implemented as an always-running plug-in with its own Scheme mini-interpreter and therefore Script-fu scripts do not use libpika or libpikaui. They interface with the PDB through the Script-fu plug-in.

Tutorials

Porting from PIKA 2 scripts

GEGL operation development

Custom data

This section list all types of data usable to enhance PIKA functionalities. If you are interested to contribute default data to PIKA, be aware that we are looking for a very good base set, not an unfinite number of data for all possible usage (even the less common ones).

Furthermore we only accept data on Libre licenses:

Of course you are free to share data usable by PIKA on any license you want on your own. Providing them as third-party PIKA extensions is probably the best idea.

Brushes

PIKA currently supports the following brush formats:

  • PIKA Brush (GBR): format to store pixmap brushes
  • PIKA Brush Pipe (GIH): format to store a series of pixmap brushes
  • PIKA Generated Brush (VBR): format of "generated" brushes
  • PIKA Brush Pixmap (GPB): OBSOLETE format to store pixel brushes
  • MyPaint brushes v1 (MYB)
  • Photoshop ABR Brush
  • Paint Shop Pro JBR Brush

We do fully support the PIKA formats obviously, as well as MyPaint brushes, since we use the official libmypaint library. We are not sure how well we support other third-party formats, especially if they had recent versions.

If you are interested in brushes from a developer perspective, you are welcome to read specifications of PIKA formats: GBR, GIH, VBR or the obsolete GPB.

If you want to contribute brushes to the official PIKA, be aware we would only accept brushes in non-obsolete PIKA formats. All these formats can be generated by PIKA itself from images.

If you want to contribute MyPaint brushes, we recommend to propose them to the MyPaint-brushes data project, which is also used by PIKA for its default MyPaint brush set.

Otherwise, you are welcome to provide brush set in any format as third-party extensions.

Dynamics

PIKA supports the PIKA Paint Dynamics format which can be generated from within PIKA.

Patterns

PIKA supports the PIKA Pattern format (PAT, whose specification is available for developers).

This format can be exported by PIKA itself.

Alternatively PIKA supports patterns from GdkPixbuf (TODO: get more information?).

Palettes

PIKA supports the PIKA Palette format which can be generated from within PIKA.

Gradients

PIKA supports the PIKA Gradient format (GGR, whose specification is available for developers) which can be generated from within PIKA.

Alternatively PIKA supports the SVG Gradient format.

Themes

GTK3 uses CSS themes. Don't be fooled though. It's not real CSS in that it doesn't have all the features of real web CSS, and since it's for desktop applications, some things are necessarily different. What it means is mostly that it "looks similar" enough that people used to web styling should not be too disorientated.

You can start by looking at the official documentation for theme migration (from GTK+2 to 3), which gives a good overview, though it's far from being perfect unfortunately.

Another good idea would be to look at existing well maintained GTK3 themes to get inspiration and see how things work.

Finally you can look at our existing themes, like the System theme. Note though that this System theme is pretty bare, and that's its goal (try to theme as few as possible over whatever is the current real system theme).

TODO: for any theme maker reading this, what we want for PIKA 3.0 are at least the following additional themes:

  • a full custom theme using neutral grayscale colors with a dark and light variant;
  • a mid-gray neutral theme.

As a last trick for theme makers, we recommend to work with the GtkInspector tool, which allows you to test CSS rules live in the CSS tab. You can run the GtkInspector by going to the File > Debug menu and selecting Start GtkInspector menu item.

It also allows you to find the name of a widget to use in your CSS rules. To do so:

  • Start the GtkInspector;
  • go on the "Objects" tab;
  • click the "target" 🞋 icon on the headerbar's top-left, then pick in PIKA interface the widget you are interested to style;
  • the widget name will be displayed on the top of the information area of the dialog.
  • Feel free to browse the various sections to see the class hierachy, CSS nodes and so on.
  • The second top-left button (just next to the target icon) allows you to switch between the details of the selected widget and the widget hierarchy (container widgets containing other widgets), which is also very useful information.

Additionally you can quickly switch between the light and dark variant of a same theme by going to "Visual" tab and switching the "Dark Variant" button ON or OFF.

Icon themes

Icon sets (a.k.a. "icon themes") have been separated from themes since PIKA 2.10 so you can have any icon theme with any theme.

We currently only support 2 such icon themes — Symbolic and Color — and we keep around the Legacy icons.

We don't want too many alternative designs as official icon themes (people are welcome to publish their favorite designs as third-party icons) though we would welcome special-purpose icon themes (e.g. high contrast).

We also welcome design updates as a whole (anyone willing to work on this should discuss with us and propose something) and obviously fixes on existing icons or adding missing icons while keeping consistent styling.

See the dedicated icons documentation for more technical information.

Tool presets

PIKA extensions (.gex)

Continuous Integration

For most of its continuous integration (macOS excepted), PIKA project uses Gitlab CI. We recommend looking the file .gitlab-ci.yml which is the startup script.

The main URL for our CI system is build.pika.org which redirects to Gitlab pipelines page.

Note that it is important to keep working CI jobs for a healthy code source. Therefore when you push some code which breaks the CI (you should receive a notification email when you do so), you are expected to look at the failed jobs' logs, try and understand the issue(s) and fix them (or ask for help). Don't just shrug this because it works locally (the point of the CI is to build in more conditions than developers usually do locally).

Of course, sometimes CI failures are out of our control, for instance when downloaded dependencies have issues, or because of runner issues. You should still check that these were reported and that packagers/maintainers of these parts are aware and working on a fix.

Automatic pipelines

At each commit pushed to the repository, several pipelines are currently running, such as:

  • Debian testing autotools and meson builds (autotools is still the official build system while meson is experimental).
  • Windows builds (cross or natively compiled).

Additionally, we test build with alternative tools or options (e.g. with Clang instead of gcc compiler) or jobs which may take much longer, such as package creation as scheduled pipelines (once every few days).

The above listing is not necessarily exhaustive nor is it meant to be. Only the .gitlab-ci.yml script is meant to be authoritative. The top comment in this file should stay as exhaustive as possible.

Manual pipelines

It is possible to trigger pipelines manually, for instance with specific jobs, if you have the "Developer" Gitlab role:

  1. go to the Pipelines page.
  2. Hit the "Run pipeline" button.
  3. Choose the branch or tag you wish to build.
  4. Add relevant variables. A list of variables named PIKA_CI_* are available (just set them to any value) and will trigger specific job lists. These variables are listed in the top comment of .gitlab-ci.yml.

Merge request pipelines

Special pipelines happen for merge request code. For instance, these also include a (non-perfect) code style check.

Additionally you can trigger Windows installer or flatpack standalone packages to be generated with the MR code as explained in gitlab-mr.md.

Release pipeline

Special pipelines happen when pushing git tags. These should be tested before a release to avoid unexpected release-time issues, as explained in our release procedure.

Exception: macOS

As an exception, macOS is currently built with the Circle-CI service. The whole CI scripts and documentation can be found in the dedicated pika-macos-build repository.

Eventually we want to move this pipeline to Gitlab as well.

Core development

When writing code, any core developer is expected to follow:

PIKA's developer wiki can also contain various valuable resources.

Finally the debugging-tips file contain many very useful tricks to help you debugging in various common cases.

Newcomers

If this is your first time contributing to PIKA, you might be interested by build instructions. The previously mentioned wiki in particular has a Hacking:Building page with various per-platform subpages. The HACKING docs will also be of interest.

You might also like to read these instructions on submitting patches.

If you are unsure what to work on, this list of bugs for newcomers might be a good start. It doesn't necessarily contain only bugs for beginner developers. Some of them might be for experienced developers who just don't know yet enough the codebase.

Nevertheless we often recommend to rather work on topics which you appreciate, or even better: fixes for bugs you encounter or features you want. These are the most self-rewarding contributions which will really make you feel like developing on PIKA means developing for yourself.

Core Contributors

Directory structure of PIKA source tree

PIKA source tree can be divided into the main application, libraries, plug-ins, data files and some stuff that don't fit into these categories. Here are the top-level directories:

Folder Description
app/ Source code of the main PIKA application
app-tools/ Source code of distributed tools
build/ Scripts for creating binary packages
cursors/ Bitmaps used to construct cursors
data/ Data files: brushes, gradients, patterns, images…
desktop/ Desktop integration files
devel-docs/ Developers documentation
docs/ Users documentation
etc/ Configuration files installed with PIKA
extensions/ Source code of extensions
icons/ Official icon themes
libpika/ Library for plug-ins (core does not link against)
libpikabase/ Basic functions shared by core and plug-ins
libpikacolor/ Color-related functions shared by core and plug-ins
libpikaconfig/ Config functions shared by core and plug-ins
libpikamath/ Mathematic operations useful for core and plug-ins
libpikamodule/ Abstracts dynamic loading of modules (used to implement loadable color selectors and display filters)
libpikathumb/ Thumbnail functions shared by core and plug-ins
libpikawidgets/ User interface elements (widgets) and utility functions shared by core and plug-ins
m4macros/ Scripts for autotools configuration
menus/ XML/XSL files used to generate menus
modules/ Color selectors and display filters loadable at run-time
pdb/ Scripts for PDB source code generation
plug-ins/ Source code for plug-ins distributed with PIKA
po/ Translations of strings used in the core application
po-libpika/ Translations of strings used in libpika
po-plug-ins/ Translations of strings used in C plug-ins
po-python/ Translations of strings used in Python plug-ins
po-script-fu/ Translations of strings used in Script-Fu scripts
po-tags/ Translations of strings used in tags
po-tips/ Translations of strings used in tips
po-windows-installer/ Translations of strings used in the Windows installer
themes/ Official themes
tools/ Source code for non-distributed PIKA-related tools
.gitlab/ Gitlab-related templates or scripts

The source code of the main PIKA application is found in the app/ directory:

Folder Description
app/actions/ Code of actions (PikaAction* defined in app/widgets/) (depends: GTK)
app/config/ Config files handling: PikaConfig interface and PikaRc object (depends: GObject)
app/core/ Core of PIKA core (depends: GObject)
app/dialogs/ Dialog widgets (depends: GTK)
app/display/ Handles displays (e.g. image windows) (depends: GTK)
app/file/ File handling routines in core (depends: GIO)
app/file-data/ PIKA file formats (gbr, gex, gih, pat) support (depends: GIO)
app/gegl/ Wrapper code for babl and GEGL API (depends: babl, GEGL)
app/gui/ Code that puts the user interface together (depends: GTK)
app/menus/ Code for menus (depends: GTK)
app/operations/ Custom GEGL operations (depends: GEGL)
app/paint/ Paint core that provides different ways to paint strokes (depends: GEGL)
app/pdb/ Core side of the Procedural Database, exposes internal functionality
app/plug-in/ Plug-in handling in core
app/propgui/ Property widgets generated from config properties (depends: GTK)
app/tests/ Core unit testing framework
app/text/ Text handling in core
app/tools/ User interface part of the tools. Actual tool functionality is in core
app/vectors/ Vectors framework in core
app/widgets/ Collection of widgets used in the application GUI
app/xcf/ XCF file handling in core

You should also check out pika-module-dependencies.svg. TODO: this SVG file is interesting yet very outdated. It should not be considered as some kind dependency rule and should be updated.

Advanced concepts

XCF

The XCF format is the core image format of PIKA, which mirrors features made available in PIKA. More than an image format, you may consider it as a work or project format, as it is not made for finale presentation of an artwork but for the work-in-progress processus.

Developers are welcome to read the specifications of XCF.

Locks

Items in an image can be locked in various ways to prevent different types of edits.

This is further explained in the specifications of locks.

UI Framework

PIKA has an evolved GUI framework, with a toolbox, dockables, menus…

This document describing how the PIKA UI framework functions and how it is implemented might be of interest.

Contexts

PIKA uses a lot a concept of "contexts". We recommend reading more about how PikaContexts are used in PIKA.

Undo

PIKA undo system can be challenging at times. This quick overview of the undo system can be of interest as a first introduction.

Parasites

PIKA has a concept of "parasite" data which basically correspond to persistent or semi-persistent data which can be attached to images or items (layers, channels, paths) within an image. These parasites are saved in the XCF format.

Parasites can also be attached globally to the PIKA session.

Parasite contents is format-free and you can use any parasite name, nevertheless PIKA itself uses parasite so you should read the descriptions of known parasites.

Metadata

PIKA supports Exif, IPTC and XMP metadata as well as various image format-specific metadata. The topic is quite huge and complex, if not overwhelming.

This old document might be of interest (or maybe not, it has not been recently reviewed and might be widely outdated; in any case, it is not a complete document at all as we definitely do a lot more nowadays). TODO: review this document and delete or update it depending of whether it still makes sense.

Tagging

Various data in PIKA can be tagged across sessions.

This document on how resource tagging in PIKA works may be of interest.