Summer of Code - Getting Started: Difference between revisions

Jump to navigation Jump to search

Summer of Code - Getting Started (edit)

Revision as of 11:00, 26 January 2023

362 bytes removed ,  26 January 2023
→‎Improve TIFF image support: removed (done in GSoC 2022)
(Strip openlibm project)
(→‎Improve TIFF image support: removed (done in GSoC 2022))
(4 intermediate revisions by 3 users not shown)
Line 1: Line 1:
{{Note|GNU Octave is a [https://summerofcode.withgoogle.com/programs/2022/organizations/gnu-octave mentoring organization for GSoC 2022].}}
{{Note|GNU Octave is applying to be a [https://summerofcode.withgoogle.com/programs/2023/organizations/gnu-octave mentoring organization for GSoC 2023].}}


Since 2011 the GNU Octave project has successfully mentored:
Since 2011 the GNU Octave project has successfully mentored:
Line 76: Line 76:
== Symbolic package ==
== Symbolic package ==


The [[Symbolic package]] provides symbolic computing and other [https://en.wikipedia.org/wiki/Computer_algebra_system computer algebra system] tools.  The main component of Symbolic is a pure m-file class "@sym" which uses the Python package [https://www.sympy.org SymPy] to do (most of) the actual computations.  The package aims to expose much of the functionality of SymPy while also providing a high level of compatibility with the Matlab Symbolic Math Toolbox.  The Symbolic package requires communication between Octave and Python.  In 2016 another GSoC project successfully re-implemented this communication using the new [[Pythonic|Pythonic package]].
The [[Symbolic package]] provides symbolic computing and other [https://en.wikipedia.org/wiki/Computer_algebra_system computer algebra system] tools.  The main component of Symbolic is a pure m-file class "@sym" which uses the Python package [https://www.sympy.org SymPy] to do (most of) the actual computations.  The package aims to expose much of the functionality of SymPy while also providing a high level of compatibility with the Matlab Symbolic Math Toolbox.  The Symbolic package requires communication between Octave and Python.  In 2016 a GSoC project successfully re-implemented this communication using the new [[Pythonic|Pythonic package]].  In 2022, a GSoC project revitalized and improved the maintainability of the package.


This project proposes to take this work further while also improving the long-term viability of the Symbolic package.  Some goals include:
Possible ideas for improving the Symbolic package:
* the possibility of using Pythonic directly rather than as one possible communication layer.  For example, we might make "@sym" a subclass of "@pyobject".  We also could stop using the "pycall_sympy__" interface and use Pythonic directly from methods.  Note: there are open questions about how to do this during a transition time when we still support other IPC mechanisms.
* exposing more functionality of SymPy with ''less glue'' in between.  For example, we could allow OO-style method calls such as <code>f.diff(x)</code> as well as <code>diff(f, x)</code>.
* exposing more functionality of SymPy with ''less glue'' in between.  For example, we could allow OO-style method calls such as <code>f.diff(x)</code> as well as <code>diff(f, x)</code>.
* Improvements to the Pythonic package and its long-term maintenance.
* Fix the storage of non-expressions by working with upstream SymPy: currently we rely on deprecated functionality in SymPy.
* fixing up Symbolic to work with the latest releases of SymPy and Octave.  The project has lagged for a few years and needs some efforts to port to recent and upcoming changes in SymPy code.
* Fix classdef-related issues in Octave itself.
* making Symbolic easier to maintain.  The project currently has a low ''bus factor'': improving the CI, making regular releases easier, improving other aspects of maintenance and making the project more welcoming to newcomers.
 
Working on this project involves and interesting and challenging mix of m-file code, Python code, and in the case of Pythonic, perhaps some lower-level C code.
 
* '''Project size''' [[#Project sizes | [?]]] and '''Difficulty'''
: ~350 hours (medium)
* '''Required skills'''
: Octave, C/C++, Python; object-oriented programming (OOP) in Octave
* '''Potential mentors'''
: [https://octave.discourse.group/u/cbm Colin B. Macdonald], [https://octave.discourse.group/u/mtmiller Mike Miller], Abhinav Tripathi
 
== Improve TIFF image support ==
 
[https://en.wikipedia.org/wiki/TIFF Tag Image File Format (TIFF)] is the de facto standard for scientific images.  Octave uses the [http://www.graphicsmagick.org/ GraphicsMagic] (GM) C++ library to handle [http://www.graphicsmagick.org/formats.html TIFF and many others image formats]. However, GM still has several limitations:
 
* GM has build option {{codeline|quantum}} which defines the bitdepth to use when reading an image:
** Building GM with '''high quantum''' means that images of smaller bitdepth will take a lot more memory when reading.
** Building GM with '''low quantum''' will make it impossible to read images of higher bitdepth. It also means that the image needs to always be rescaled to the correct range.
* GM supports unsigned integers only, thus incorrectly reading files such as TIFF with floating-point data.
* GM hides details of the image such as whether the image file is indexed.  This makes it hard to access the real data stored on file.
 
This project aims to implement better TIFF image support using [https://en.wikipedia.org/wiki/Libtiff libtiff], while leaving GM handle all other image formats.  After writing a [https://octave.org/doc/v6.1.0/classdef-Classes.html classdef] interface to libtiff, improve the Octave functions {{manual|imread}}, {{manual|imwrite}}, and {{manual|imfinfo}} to make use of it.
 
* '''Project size''' [[#Project sizes | [?]]] and '''Difficulty'''
: ~175 hours (medium)
* '''Required skills'''
: Octave, C/C++
* '''Potential mentors'''
: [https://octave.discourse.group/u/carandraug Carnë Draug]


== PolarAxes and Plotting Improvements ==
== PolarAxes and Plotting Improvements ==
Line 169: Line 139:
* '''Potential mentors'''
* '''Potential mentors'''
: [https://octave.discourse.group/u/kakila KaKiLa]
: [https://octave.discourse.group/u/kakila KaKiLa]
== Better tab completion ==
Links: [https://savannah.gnu.org/bugs/index.php?62492 https://savannah.gnu.org/bugs/index.php?62492] and [https://savannah.gnu.org/bugs/?53384 https://savannah.gnu.org/bugs/?53384]
Description: currently pressing Tab at the Octave command prompt attempts autocompletion of all identifiers in scope (variables, functions, classdefs etc) as well as files and directories in the path. It is not context sensitive.
Project: Improve tab completion. For example,
* Typing
: load x
and then pressing tab should ideally give all loadable files and directories starting with x, not unrelated results like variables or functions.
* Typing
:cd
and tab should only give directories.
* Any file and directory names that are included in the results should include those with spaces and special characters including those that would be interpreted as operators by Octave.
* Typing commands like
: axis
or
: format
and pressing tab should give only those options relevant to that command. E.g. format can be followed by short / long / loose / compact etc but not by a file or variable. Similarly axis can be followed by ij / xy / tight / equal / actual limits etc but not by files or directories. And so on for other commands. This should be made possible for both preexisting commands and for yet-to-be-written commands without any rewriting of existing function code or documentation.
To get more examples, see how bash completion works. You can type git or hg and then tab and it will give the list of available commands. If you type "sort --r" and then tab, it gives the list of options to sort starting with "--r", etc.
== Graphics rendering back to front sorting ==
Several incompatibilities have been identified in how Octave plots transparent objects in 3D, causing certain transparent objects to hide opaque objects behind them even though they're not supposed to. The vast majority of them were isolated to one problem: if the objects to be drawn are rendered such that the one farthest away from the viewer is rendered first and nearer objects are rendered on top of that, then transparency would be automatically achieved, but this needs very careful coding to stay performant and to avoid rendering objects that will be overwritten fully by others. See [https://savannah.gnu.org/bugs/?57980] for a summary.
This project can be a GSoC for a student who understands 3D graphics rendering.


= Project sizes =
= Project sizes =
Retrieved from "https://wiki.octave.org/Special:MobileDiff/14466...14597"

Navigation menu