Google Summer of Code 2016 Ideas

From ReactOS Wiki
Revision as of 04:44, 3 March 2016 by Colin Finck (talk | contribs) (Added the TreeView idea)
Jump to: navigation, search

This page lists all project ideas for our Google Summer of Code 2016 application. Students should also visit our general GSoC 2016 page for more information including a Student Guide.

Your idea!

We are eager to hear about your proposal to improve either ReactOS or its infrastructure. Come and propose your project on ReactOS developers mailing-list or on IRC, to get feedback about its feasibility. If your project is doable, we will be glad to provide you a mentor so that you can succeed in your project.

Don't forget that ReactOS is a big project and application fields are wide. You can choose between user-mode, kernel-mode development. But also between application, or dll/driver development. Or between working on ReactOS itself or on extra tools used by developers for their daily work on ReactOS. And in the end, if all you prefer is web development, we also have needs there.

Skills needed
Yours
Difficulty

The one you'll have set

Extras

You'll be working on your own project and be able to choose its direction along with your mentor :-).

Drivers

AHCI SATA Storage Driver

Implement a Windows-compatible AHCI driver (msahci.sys) for the NT 5.2 architecture. This driver would be unique in that it would bring a feature of Vista back to prior Windows versions and could be used both in ReactOS and Windows. Nearly all new machines feature AHCI controllers exclusively and the ability to fallback to IDE emulation will likely disappear over the coming years. Therefore, an AHCI driver is necessary for the project to ensure it can run on real machines in the present and future. This driver would be able to provide AHCI support for Windows versions before Vista like 2000, XP, 2003, and obviously ReactOS. This driver would also provide a fantastic test case for our SCSI Port driver which has nothing testing its PnP path as of now.

References
SCSI Port API: http://msdn.microsoft.com/en-us/library/windows/hardware/ff565353(v=vs.85).aspx
AHCI specification: http://www.intel.com/content/www/us/en/io/serial-ata/ahci.html
Universal ATA driver with PATA/SATA/AHCI support (opensource): http://alter.org.ua/soft/win/uni_ata/
Skills needed
Kernel-mode development
Familiarity with SCSI Port APIs


Intel High Definition Audio Bus Driver

Implement support for the Intel High Definition Audio specification for sound cards.

Benefits
Modern audio cards use the Intel High Definition Audio specification. Therefore they rely on a bus driver to communicate with the audio hardware. The goal is to write a bus driver which supports those new cards. The bus driver needs to work on a Microsoft NT system and with ReactOS.
References
HD Audio Device Driver Interface whitepaper: http://msdn.microsoft.com/en-us/windows/hardware/gg462966
Programming the HD Audio DDI: http://msdn.microsoft.com/en-us/library/ff536442%28v=vs.85%29.aspx
HD Audio Bus Drivers: http://msdn.microsoft.com/en-us/library/ff536434%28v=vs.85%29.aspx
Skills Needed
Kernel driver programming experience
Bus driver programming experience
Familiarity with Intel HD Audio specification
Audio kernel streaming


Fully Integrate lwIP

Useful for: End-users

Currently ReactOS uses two entirely separate code bases for its network operations, UDP utilizes lwIP while TCP uses much older code. While UDP runs well albeit slowly, TCP is a minor mess and has major bugs that prevent applications like Chrome from working. lwIP has a working TCP stack, along with support for a few other network interfaces that we either have poor or non-existent support for. Fully integrating lwIP into ReactOS, especially the newer versions which have support for things like multithreading, would significantly improve ReactOS' usability.

Benefits
Better source code
Better compatibility and/or performance
Needed Skills
Windows Networking stack knowledge
TCP/IP protocol tech knowledge
Resources
VM image for Windows 2003 to test code on: https://www.microsoft.com/en-us/download/details.aspx?id=19727
Extra

Make the performance of a new tcpip.sys in Windows better than its native driver. Ambitious, but fun.

NT "Wine audio" driver

Useful for: End-users

Currently, ReactOS imports mmdevapi DLL from Wine, but it doesn't have a "driver" for it. In Wine, the driver is an implementation of various COM interfaces on top of libraries such as ALSA or OSS. However, it is perfectly possible to implement such a driver for Windows NT (preferably Vista+) using audio IOCTL to directly talk to the windows device.

Benefits
Vista+ compatible sound stack (our portcls.sys is already there)
Importing the latest Wine audio related libraries becomes possible (winmm, mmdevapi, dsound, etc.)
One place to rule them all: the only direct interface between user mode and driver.
References
WINE documentation about its Sound System
Core Audio
Wine implementation on top of alsa
Needed Skills
Windows Audio Stack knowledge
COM knowledge


NTFS driver improvements

Useful for: End-users and developers

During the last spring, NTFS support was finally added in ReactOS. If it was a major step forward, it still requires huge amount of work. It was implemented without caching support, and some features of NTFS are also totally ignored. We are looking for students willing to discover one of the most used file system and to improve its supports in ReactOS. There are various options offered to students: either work on improvements for RO support and/or start implementing RW features.

Benefits
Allowing ReactOS to interact with volumes managed in recent Windows
Perhaps allowing ReactOS to have a journaled system
References
Various documents on the Internet (including MSDN)
Windows Internals books
NTFS-3G FUSE driver
Needed Skills
Kernel world knowledge
FUSE dev knowledge might be a plus


Plug and Play (PnP) Storage Stack

Right now, PnP support in the storage stack is close to be there. mountmgr.sys has already been implemented, and recent tests show that plugging USB sticks work. However, "traditional" storage stack is not PnP-aware, meaning that hot-plugging SATA or SCSI device is not possible. Alleviating PnP support in scsiport would be the first task of this project. Then, turning uniata into a PnP-aware driver would permit SATA devices hot-plugging. Finally, PnP-aware "cdrom_new" and "disk_new" drivers can be activated.

References
SCSI Port API
Windows Open-Source Storage Driver Samples based on the PnP Storage Stack
Universal ATA driver with PATA/SATA/AHCI support (opensource)
Skills needed
Software Development in C and under Windows
Experience with developing low-level Kernel-Mode hardware drivers
Previous experience with Windows Storage APIs would be a huge plus


USB 3.0 Support

USB 3.0 (Also known as eXtensible Host Controller Interface or xHCI) is a computer interface specification that defines a register-level description of a Host Controller for Universal Serial bus (USB) which is capable of interfacing to USB 1.x, 2.0, and 3.0 compatible devices. It's increasingly present these days. Having a USB 3.0 stack would make ReactOS available to the newest external device interface among others.

References
wikipedia:Extensible Host Controller Interface has links to the required specifications
Current OHCI and EHCI implementations
Needed Skills
Software Development in C and under Windows
Experience with developing low-level Kernel-Mode hardware drivers
Previous experience with USB-related development would be a huge plus

Networking

Integrating SMB into ReactOS

Useful for: End-users and developers

SMB/CIFS is omnipresent for file sharing under Windows. It's one of the features new users to ReactOS always expect and then cannot find anywhere.

As such, ReactOS finally needs an implementation of SMB properly integrated into the operating system. SMB heavily relies on RPC and UNC pathes and previous work in both areas exists. A potential developer should base the work on the mature Samba Open-Source implementation of SMB. However, Samba is developed for UNIX systems and therefore parts of its code require careful porting to create a native implementation for ReactOS.

Benefits

Having SMB would highly improve ReactOS' abilities to transfer files over the network, both for users and developers! It would also lay the ground for other Windows network services such as Printer Sharing.

References
Samba Wiki
MS-SMB Server Message Block (SMB) Protocol Open Specification (and related ones on this site)
Needed Skills
Software Development in C and under Windows
Familiarity with Win32 APIs and Networking
Experience with glueing code of different projects together


SSH Service

Useful for: End-users and developers

Implement a native Windows service for accepting SSH connections and authenticating with Windows accounts to allow the user to interact with Windows through the command prompt should they not need access to the GUI shell. Once authenticated, any commands passed through SSH are executed using the user's credentials to prevent abuse of the service's privileges. And as the user is not technically logged in, there are no restrictions with respect to how many connections are permitted except as system resources allow.

Benefits
Besides cygwin, few options are available for users who wish to remote into a Windows machine for simple command line access. Cygwin is also horrendously intrusive on Windows systems. This implementation would provide a free SSH server implementation for Windows that was designed to integrate cleanly and present users with access to the Windows command shell, not try and lie to the user and pretend they are in a Unix environment. As a remote administration tool, this opens Windows systems to considerably more flexibility for connecting to and managing them. Administrators on Linux systems will also not need to relax security settings on Windows in order to use Linux RDP clients.
Skills Needed
Network development experience.
Knowledge/familiarity with SSH protocol.


Terminal Services

Useful for: End-users and developers

Implement support for terminal services, allowing inbound remote desktop connections to ReactOS. This encompasses implementation of input and video drivers to handle output over the network.

The terminal services system provides functionality for securely connecting remote clients and servers, for channeling communication between components of remote clients and servers, and for managing servers. It implements the Remote Desktop Protocol (RDP) which is a multi- channel protocol that allows users of a remote client to connect to a server over a network.

Benefits
Terminal services support would allow ReactOS to service as a terminal server/thin client server. Organizations that wish to provide a shared workstation with Windows would have a free alternative that does not have complex licensing terms covering multi-session usage. The display driver for terminal services can also be used to support fast user-switching and even possibly multi-monitor support.
References
ReactOS Terminal Services
Skills Needed
Network development experience.
Driver development experience.
Knowledge/familiarity with RDP protocol.


Durability

Kernel mode test suite

Useful for: Developers

Improve our existing kernel mode test suite (created by a previous successful GSoC student) by adding comprehensive new tests in areas previously untouched by the test suite like the kernel caching APIs and PnP. The goal is to extensively test Native API functions exported by the kernel. Reference behavior could be checked by running the test suite on original Windows operating system.

Benefits
Improved system stability.
More information about undocumented behavior.
Improved compatibility with third party drivers.
Skills Needed
Native API knowledge
NT driver development skills


Win32k test suite

Useful for: Developers

Right now we have basically no tests that cover the win32k module. As win32k is the cornerstone of the interface between the Win32 Subsystem and the display drivers, lack of tests leads to inconsistency, guess work and frustration when it comes to improving driver compatibility. A virtual display driver (à la kmtest.sys) allows us to test the functionality and behavior of win32k. If time permits, it would also be possible to add a mechanism to test videoprt.sys functionality, and the relationship between a display driver and its miniport counterpart.

References
Windows 2000 display driver model reference
Benefits
Improved stability.
More information about undocumented behavior.
Improved compatibility with third party drivers.
Needed skills
Display drivers development experience
Win32 API knowledge


System Enhancements

Audio Mixer

Implement support for mixing of audio streams.

Benefits
Audio mixer is responsible for routing of multiple audio streams. This would be very beneficial to ReactOS as audio driver support has improved yet the use of these drivers is limited by the lack of an Audio Mixer. In the end of the project multiple audio streams should be able to be played at the same time.
References
Audio Mixer Reference
DirectKS Sample Application download
Programming Services
"Secret Rabbit Code" (SRC) Sample Rate Converter aka libsamplerate
Skills Needed
Windows Services familiarity
Basic Audio file format familiarity
Audio mixing algorithms / libraries familiarity
Basic kernel streaming familiarity


Folder TreeView in Explorer

Our current Explorer still lacks a Folder TreeView on the left. This one should show a combination of real folders on disk and virtual Shell Folders (e.g. "Control Panel") just like Windows does. It must be implemented compatible to the Windows Explorer interfaces to maintain compatibility with existing applications relying on it.

Benefits
Such a TreeView would highly improve navigating through the file system and make the ReactOS Explorer feel like a powerful file management tool.
Skills Needed
Development in C and under Windows
Knowledge of COM will be required at some point during development.


Performance Data Registry

Access to performance data on Windows is done primarily through the registry API, accessing something known as the performance data hive. This hive does not exist as a genuine file but is in reality a collection of data exported by various OS components, drivers, services, and even applications. Many of the performance values provided through the performance data registry is not available in any other form. The absence of support for performance counters renders many diagnostic utilities from Microsoft broken and is also an impediment to application compatibility. Condor is an example of a third party application that uses the performance data registry for process and resource usage tracking.

Benefits
Besides application compatibility, the performance data registry is one of the most difficult to use public interfaces in Windows. The layout of its data structures makes querying and accessing values a highly manual process. Documentation produced from this effort would provide better guidelines for third parties to access the performance data registry and better use the information published by the system and Microsoft's own applications such as the .NET runtime or the IIS service.
References
Description of performance counters and types in Windows 2003: http://technet.microsoft.com/en-us/library/cc776490%28WS.10%29.aspx
Performance counter data structures: http://msdn.microsoft.com/en-us/library/aa373093%28v=VS.85%29.aspx
Example code for accessing performance data through registry: http://msdn.microsoft.com/en-us/library/aa373219%28v=VS.85%29.aspx
Skills Needed
Registry API familiarity.
Performance registry structure familiarity.
Performance counter familiarity.


Security Controls

Implement support for user permissions based on Access Control Lists.

Benefits
This will allow separation of users into groups and better control over user privileges in ReactOS. Security controls will then allow supporting multiple users on the same system.
Skills Needed
Familiarity with Windows security and permissions models.


Management Console

Implement Microsoft Management Console (MMC). The MMC provides an interface for various management tools, both from Microsoft and third parties, called snap-ins. These snap-ins are standalone programs dynamically loaded into an MMC console to perform a specific configuration task, such as configuring a network or managing disk drives.

Benefits
Easier snap-ins development.
Consistent user interface of management tools.
One configurable place to access key management and configuration apps.
References
What is MMC
Developing for MMC 3.0
Skills Needed
Basic Windows development experience.
Knowledge of COM.


Applications Manager Rapps

Useful for: End-users

RAPPS is a lightweight GUI utility which downloads and installs various popular apps and redistributable packages in Windows and ReactOS. It can be significantly improved to become a really useful tool for not only for ReactOS but for Windows users also who want to manage their installed applications like Steam manages games installation/updating.

A list of features to be implemented:

Adding command line parameters support to RAPPS. For example: "rapps install chrome" (just like your favourite Linux distro package manager) or "rapps remove msvc2010_redist"
Separate download and install, so that a user could choose just to download the apps into a local cache and later install some or all of them (just like Steam does Download, then it asks to Install, or NVidia driver updaters for example)
Automatically "install" simple apps by unpacking the archive they come in and execute a script inside it (if any) which creates shortcuts on the desktop
Skills needed
WinAPI
C language
Difficulty

Medium

Extras

A chance to work on an app potentially used by many people throughout the world


GUI 1st Stage Installer

Useful for: End-users

To make ReactOS more user friendly for installation, a GUI installation instead of a rather limited text-mode installer is needed. This could be run off of the LiveCD and act as a frontend to the disk formatter and installer we already have implemented.

Benefits
Having a GUI installer would allow the project to merge the bootcd and LiveCD and ensure that the LiveCD gets more testing to make sure it is not broken and simply neglected. A GUI installer would also be less intimidating to new users, especially those used to Vista or 7's installer.
Skills Needed
Win32 API familiarity.
Partitioning and setup experience.

Win32 Subsystem

Multi-Monitor Support

The implementation of multi-monitor support is until now almost inexistent. Display Device Drivers exposes them thanks to the HwVidGetVideoChildDescriptor callback, which videoprt.sys uses to enumerate monitors attached to it to then pass the information to win32k.sys. It is then the role of win32k.sys to make the link between the two (or more) monitors, either cloning or extending the desktop and exposing the relevant features to client applications.

Benefits
Multi-monitor support would bring ReactOS into a serious player in professional desktop applications, given the increasing number of double-screen installations present in modern open-spaces. Document comparison, permanently visible e-mail reader or presentation applications are just a few benefits professionals get when using multiple monitor.l
Multi-Monitor support would also find its place in the casual computing experience. Writing e-mails while watching a movie, or chatting with friends about current football play is the things people do in XXIst century!
From a technical stance, properly supporting this feature would increase driver support, and WIN32 application compatibility regarding APIs such as EnumDisplaySettings.
References
http://msdn.microsoft.com/en-us/library/windows/desktop/dd145071%28v=vs.85%29.aspx
http://msdn.microsoft.com/en-us/library/ff569172%28v=VS.85%29.aspx
http://www.vesa.org/
Virtualbox video driver source
Skills Needed
Experience with Windows display drivers.
Experience with WINAPI, especially gdi32/user32.: Experience with PNP, as this feature would require detection of connected monitors: Experience with EDID/VESA standards.


Enhancing web presence of ReactOS

Implement a REST API in testman

Useful for: Developers

For its development, ReactOS project has been using with success for years a complete tools set. Each time a commit is made to our trunk, our buildbot infrastructure (https://build.reactos.org) is in charge of rebuilding trunk and starting a bunch of tests (circa 10M tests) on it to make sure not regressions came in. When these tests are done, they are uploaded to testman (https://www.reactos.org/testman), which makes them properly readable and usable for developers. Over years, the question of being able to catch more or less automatically regressions patterns was raised. Also, regtests provide a good performance indicator for ReactOS and evaluating their performances over commits can be critical as well.

There is then a need to be able to directly query Testman via an API to gather the raw data (tests, time to perform then) so that more tools can make use of these results. The student would then have the responsibility to implement such an API into testman directly.

Skills needed
PHP/MySQL development
REST API development
Difficulty

The project in itself shouldn't be difficult. Testman is a well-known product develop in-house. But a deep attention will have to be set on security and the performances of the API.

Extras

If the student goes fast enough, we will be able to extend this project, by for instance looking at how to develop a skeleton application to make use of the API. Or to directly develop a complete application to evaluate the performances of the tests over time.