[Scummvm-devel] GSoC 2009, release plans

Max Horn max at quendi.de
Thu Jan 8 00:13:10 CET 2009


Am 07.01.2009 um 19:15 schrieb Sven Hesse:

> On 2009-01-07 19:54:28 +0200, Eugene Sandulenko wrote:
>> Hi Team,
>>
>> Google confirmed that there will be GSoC in 2009, so it is really  
>> time
>> to think on the ideas to do. Please, start doing it already now so we
>> will be well prepared in case we will be accepted.
>
> Would putting up TFMX support again be a good idea? It is kinda sad
> it didn't really work out last year...

We could put it up. But personally, I am not sure it is *that*  
important... Oh, and BTW, I'd love if we could complete some more of  
the past projects. Like, the virtual keyboard stuff still needs to be  
merged...

Anyway, as for project ideas, here are two:

1) Implement support for loadable modules for NDS and/or other (ARM  
based?) ports, following the example set by Marcus' loadable module  
code for the Dreamcast.

2) Conversion of our build system to CMake. This would allow us to  
generate MS VC, XCode, Eclipse and other IDE project files, just as  
well as regular Makefiles. It also includes such fancy things as good  
integration with cross compilation environments, building of universal  
Mac binaries, etc.

Combined with that, adding support for building engine plugins  
"outside" the ScummVM tree, to make thingseasier for 3rd party engine  
developers.


Besides this, I also think we should put a lot more effort into  
selecting students -- and consider asking for fewer students. Some  
suggestions

* GSoC is a _full_time_ job!! Make sure all students know that. Be  
*very* explicit about this, even if we might risk loosing some (but  
good ridance to those!)
* conduct IRC interviews!
* conduct phone interviews?
* require sample patch to be commited -- something trivial (fix a  
typo), forcing them to use our patch tracker and the "diff" tool
* require detailed & complete list of other activities during summer - 
 > side jobs? uni task (lectures, exams, ... ) ? vacation, travel?  
etc.. With the explicit notice that "forgetting" to list some may lead  
to immediate abortion of the project

And during GSoC, I think we should add the following hard rule: If  
gone w/o any message / status update for more than N days -> automatic  
fail! Of course we need tell all students about this rule in advance,  
and we'd make exceptions if we receive a notification by the student  
*in advance* that he will be gone, with valid reasons etc.


Yeah, we may risk loosing some by being so hard. But I feel we had  
some really bad experiences in the past. From the perspective of what  
ScummVM gained, sadly I feel that to many of our GSoC projects in 2007  
and 2008 were effectively failures :/. Although it improved a lot in  
2008

Success 2007:
1) File system abstraction layer
2) Improve AGI engine
3) Tools: Write a portable GUI for the tools (???? well we have code  
in there, but I wouldn't really call it finished nor polished :/)

Success 2008:
4) Return to Launcher and Global Main Menu / Savestate Management
5) Graphical User Interface overhaul for ScummVM
6) Virtual Keyboard and Keymapper (??? I still have hopes, but it is  
not actually there...)
7) Adding support for Operation Stealth (as far as I can tell at least).

Failure 2007:
1) Small devices backend
2) Mixer improvements
3) Revive ScummEX
4) Tools: Game script (bytecode) decompiler

Failure 2008:
5) FreeSCI Engine
6) Support for AMIGA Audio Formats: TFMX and MaxTrax



Bye,
Max





More information about the Scummvm-devel mailing list