Re: [Gnash] Gnash roadmap ?

[Rob Savoye]

Nicolas Cannasse wrote:

  I'm back from my climbing trip, and will try to work my way through 
the over 200 messages I got about Gnash while I was gone.

> I would like to hear more about Gnash and know in which area some work 
> is needed. I had a quick look at the sources, some parts are looking a 
> bit messy right now :) The code seems to come from GameSWF and maybe 
> other parts. Is there some kind of "redesign from scratch" (using 
> previous code but rearchitectured) planned ? What is the current status 
> and roadmap ?

  Gnash is based on GameSWF, which I also worked on. A lot of the 
GameSWF code is quite messy, but was suitable for Thatcher's purposes. 
(I appreciate all the work he did, Gnash wouldn't exist without it). 
Once the plugin works, I have definite plans to do a major code cleanup 
to the old GameSWF code, for one thing using real classes instead of 
structs. I'd also like to break some of the huge files apart into 
smaller, more modular ones the way I did the XML AS classes.

  There is no roadmap as of yet, I only did the fork a few weeks ago. 
:-) A short roadmap would probably be:

        1. Get the plugin working
        2. Write a real roadmap and task list
        3. Add much more to the manual
        4. Do some testing and list out what isn't implemented yet
        5. Develop a test suite, and implement what's left
        6. Start a major code cleanup
        7. Look for performance improvements and do them
        8. Write a non-OpenGL backend for embedded devices
        9. SWF 8 support

  This looks like a short list, but will involve a lot of work. This is 
where some of us working together can accomplish these tasks much more 
quickly than if I have to do them all myself. Most of the work on Gnash 
at this point was to fork GameSWF into a more suitable framework, and to 
get all the initial project infrastructure created. (the web sites, CVS, 
etc...)

> Also, one of my concerns is about the usage of OpenGL. While this might 
> speed up things on the desktop, it would be interesting to comeup with a 

  I have a strong interest in having Gnash run on embedded devices 
without OpenGL or any hardware acceleration support. I was debating 
using Cairo, but at this point the API to use for this backend is open 
to discussion.

  This is one of the prime motivators for me to do Gnash. Not just 
because having a free plugin would be cool, but to have the standalone 
player work on my Zaurus, or other small footprint devices. As more and 
more people and projects want to use Flash, the free software world 
needs a truly portable player.

> I'm interested also about the current AS VM in Gnash. How does it 
> compete with Macromedia Flash7 VM ? (which is awfully slow) Is there 
> some plans in the future to support Flash8 features ? (filters might not 
> be a so good idea with OpenGL ?) What about the Flash 8.5 virtual 
> machine which defines a new set of bytecode - currently being reverse 
> engineered little-by-little ? I know it's a bit early to ask about all 
> theses but they might be important in the way the code is designed from 
> the beginning.

  I haven't used the MacroMedia tools, so I can't really offer any 
comparison. I'd have to look at some clean documentation (to not 
contaminate Gnash) for SWF 8. I don't see a huge issue with changing 
byte codes, but my guess is other things have changed as well that could 
be a big issue. I won't know till I dig into this deeper.

  Currently most Flash movies I've been collecting don't use SWF 8 at 
all, so initially I'd like to make the SWF 7 format (and earlier) work 
well enough as a plugin to play most movies before launching into 
implementing SWF 8. I would like to have a good ideas as to the SWF 
structure though, as when I start on the code rearranging of the 
interpreter, I can add hooks to make it easier to keep Gnash in sync 
with the evolving Flash spec.

  So at this point I'm asking for more developers so we can make Gnash 
a fully functional player and plugin. If the free software world can 
handle big projects like Firefox, OpenOffice, or the Linux kernel, we 
should be able to handle a Flash player. :-)

        - rob -