This page lists ideas for projects which GDB would like to have implemented by GSoC students. We may also consider items which are not in this list. Please discuss your chosen idea with the GDB community before applying for the GSoC. You can find us in the #gdb channel at Freenode, but we prefer that GSoC discussions happen in the GDB mailing list, namely gdb [at]

Students can work on GDB ideas through the GNU Project organization. Please read the GNU guidelines for Summer of Code projects, it has good tips which will help you submit a proposal with more chances of being successful.


There are some ideas for the GDB project in the ProjectIdeas page. This page was initially created with a few ideas that we came up during a very quick discussion, but you are totally free to propose new ideas if you wish.

If you are seriously considering and preparing to work on a GSoC project in GDB, read the DeveloperTips, the GDB Internals document (very incomplete, but gives an idea on how GDB is organized internally) and the CONTRIBUTE document in <gdb-src>/gdb/CONTRIBUTE.

Console interface to GDB by MI protocol

GDB currently has code to communicate with the world in two ways:

It is worth noting that some interfaces (DDD, TUI mode inside GDB, or older Emacs interfaces) do not use MI.

This makes GDB code complicated and error-prone, because one has to be careful about what to output depending on which mode is being used. It would be cleaner to make the CLI separated on top of the MI, exactly like other (graphical) interfaces are currently doing. The idea is that GDB code should deal with functionality, and not the output format. Currently, the MI protocol has some functionality flaws which cannot be easily fixed without breaking compatibility with existing front-ends.

The idea is to investigate and continue Jan Kratochvil's work-in-progress (see the branch here).

ARI revamp

GDB has a subproject called ARI, which is intended to catch problems with indentation, coding style, usage of deprecated features/functions, etc. It is mostly written using shell and awk, and you can see an instance of it here. It has many limitations, and needs to be improved (maybe rewritten). A good task would be to take this job (with the blessing from Pierre, the ARI maintainer!) and see how the ARI can be improved, and how to integrate it with make check.

None: SummerOfCode2014/Ideas (last edited 2014-02-19 23:05:53 by SergioDuriganJr)

All content (C) 2008 Free Software Foundation. For terms of use, redistribution, and modification, please see the WikiLicense page.