Re: [anjuta-devel] GSoC: Anjuta as an AVR development environment

[Lucas van Dijk <info return1 net>]

Thanks for your response :)

That was indeed my biggest concern, embedded development isn't a really big audience, and Anjuta isn't directly targeting at them right now. But what doesn't exists can always come right? :)

About your comment getting the three months full, I think it's enough stuff to get me busy for three months, especially when you add assembly support. The project wizard and avr-gcc won't be too much work indeed, but the debugging part is a little bit bigger. Showing the pin statuses in a dockable widget, reading/writing EEPROM and RAM, and some configuration panels in the project properties dialog, although that's not that much work as I saw a lot of XML UI files and little code.

If I really have time left, I will be writing a scintilla lexer for AVR assembly, write a nice window to view the compiled assembly, I don't think I have trouble filling those three months :)

I will look into that bug/feature request, I think it's indeed a nice small project to understand the Anjuta architecture a bit better.

Regards,
Lucas

On Tue, Mar 22, 2011 at 4:50 AM, Johannes Schmid <jhs jsschmid de> wrote:

Hi!

First, thanks for your interest in the Anjuta project. I want to be
honest that I don't think AVR integration as a high priority even if I
work a lot with these Chips (mostly with AVR Studio) but it still can
be interesting and maybe someone else of the developers is willing to
mentor it.

At least, the only other option I know on Linux is Eclipse and it is
pretty big and kind of hard to setup.

> My idea is to add some AVR specific functionality to Anjuta. GCC
> already has an AVR version, so does GDB. I would like to create
> project/makefile templates to easily create projects meant to be
> programmed on an AVR microcontroller. Also integrate avr-gcc and
> avr-gdb to make compiling and debugging a piece of cake.

OK, I think cross-compiling is pretty trivial. You would just have to
setup a project-wizard template for anjuta that add the required options
the the build system. You might want to try that before GSoC, it should
be easy.

> Several GDB servers exists which are useful for AVR development. One
> of them is simulavr [4_], as the name suggests, a simulator for some
> types of AVR microcontrollers. And there's also avarice [5_], a GDB
> server for on chip debugging using JTAG, a very common protocol used
> for on chip debugging. Adding an option to the project creation wizard
> to specify which type of AVR microcontroller, and also an option if
> they want to use the simulator or on chip debugging would be very nice
> to have, and makes things really easy.

I think JTAG is basically the standard if people are debugging on the
AVR sp avarice seems the way to go. I am not really sure how a simulator
would fit into anjuta but you might want to point that out. Having
possibilites to show register content and pin status could be nice.
Anyway it would be cool to have remote debugging support in general, not
only limited to avr architecture.

> When the program has been compiled, it should be transferred to the
> AVR microcontroller itself. There's a tool called avrdude which
> handles programming the AVR. A nice frontend for this program should
> also be included.

That's also rather easy I think. It is basically one dialog with some
options for avrdude and a bit of anjuta-launcher glue code to execute
the command.

So overall it feels to me that this might not be enough for 3 months
full-time on the other hand we have badly overestimated the time in the
last years.
You might compete with other ideas here so it is important that we can
see that you are really into this which is best done by participating
the development before GSoC. There is a bug in bugzilla about being able
to set the gdb path with touches areas you might need in your project:
https://bugzilla.gnome.org/show_bug.cgi?id=645081
You might want to have a look at that.

So again, thanks for your interest, I hope my words don't sound too
discouraging overall as I really like when people want to contribute to
the project. Until now, anjuta just got into a really different
direction as targeting embedding systems and I am not entirely sure how
many resources to put into this.

Regards,
Johannes