The seconds part of the roadmap is tricky. Basically this is the main reason why I have to write a new version of the player, instead of continuing the original one. What this point is? This was the intention for the original version:

Simple, non-bloated interface. The interface must be intuitive to navigate and provide fast access to everything one needs.

Customizable. One must be able to customize the player. The current player is very customizable, but it is not done the right way. The console approach is not too user friendly.

And what I wanted for the new version:

Eye candy. I am not a big lover of skins but they could be nice if done right. Also there is always the possibility of making Windows like skin.

My first consideration is about the first point and the eye candy. I decided to use HTMLayout as interface library because it is very fast and allows for visually appealing interface. But it seems that most people like the old school interface. In fact I like it too. In my opinion it does all what it should do. Provide fast access to all functions. So I have second thoughts about HTMLayout. Maybe I should just use DWT, as in the original version? Now there is new version of DWT that is also available on Linux. The downside of this is that it is a bit slower than the old DWT and than HTMLayout. So the options are:

DWT:

Pros

1. Will not require full rewrite as some of the old code could be reused
2. Available for both Linux and Windows
3. Keeps the old-school looks

Cons

1. Is a bit slower
2. Some OS specific things may need to be implemented in a hacky way, i.e. unclean code
3. Keeps the old-school looks - nobody complains
   
4. Less customizable (including from plugin’s point of view) - solved
   

HTMLayout:

Pros

1. Very fast
2. Allows web-style interface, skinning, innovative stuff
3. More customizable (including from plugin’s point of view)

Cons

1. Requires a full rewrite
2. It is harder to integrate MPlayer inside it
3. Windows only at the moment - will change in near future
   

Tough choice. Now the second point is how to make the player customizable without bloated interface and preferably without dialog boxes. It seems that the console is not too good approach. Eventually it became not very useful even for me, because I forgot how to use it. If using HTMLayout some innovative way of configuration could be introduced. But DWT is pretty limited - endless menus or dialog boxes. I don’t like both. Love to hear other opinions.

Update:

I failed to integrate MPlayer in HTMLayout in an efficient manner so I am back to DWT. I’ve been working on it for some days and I have some fresh ideas how to make it more flexible.

Blazing performance. Both startup and playing the movie (this means fast rewinding). The next version must be even faster.

This one seems easy. MPlayer is already very fast and HTMLayout is blazing fast on startup and redrawing the interface so we have part of the work done already. But still there is one thing. Drawing the movie on the screen. I.e. the video output driver. Which one to use? MPlayer comes with severals Windows compatible drivers - DirectX, OpenGL and OpenGL 2 (this is probably just the name, not referring to OpenGL 2.0).

DirectX - should work pretty fast everywhere, supports hardware acceleration and works on Vista (I guess). Does not support pan-scan and resizes the OSD the wrong way. I.e. the subtitles are drawn on the movie with it’s original size and then the whole thing is resized (to fullscreen for example). This results uglier subtitles for movies with small resolution. Also uses hardware overlays for better performance, which has some side effects, not such a big problem though.

OpenGL - generally fast but is limited by the hardware. The video card needs to support some features. In this version of the driver, the maximum video resolution is limited by the maximum supported texture size. And 1280 pixels is pretty common video size, and not so common texture size… Supports only texture sizes that are power of two. Otherwise they are not hardware accelerated. And software OpenGL is much slower than software non-OpenGL. Have you tried playing Half-Life with OpenGL on S3 1MB video card :) ? Try software OpenGL even on a modern video card… It is very slow; This is the only driver that resizes the movie and then renders the OSD. This results very crisp subtitles. Really nice! Also supports pan-scan.

OpenGL 2 - doesn’t have the above limitations but the OSD is rendered the ugly way. What a pity. As far as I know Microsoft crippled OpenGL big time in Vista.

Then which one to choose? There is something wrong in each one of these drivers… So I will write my own “rules” driver. The plan is to use shared memory (i.e. MPlayer and rulesPlayer share the same memory location; this saves lengthy buffer copying). Let MPlayer render its stuff directly into the memory buffer -> notify rulesPlayer -> render the buffer in the window. This seems fast enough to me, but who knows. We are missing any hardware acceleration this way. But once you have the whole video frame you can crop it, you can pan-scan it, whatever. I even want to make HTMLayout render the subtitles instead of MPlayer. Don’t worry, it supports antialiased text with shadows that are not less nice than MPlayer’s. The only thing that bothers me is the performance. But we will see about that soon… If it doesn’t work out so well I will try to fix the issues with above drivers and let the user choose the one which works best on his hardware.

As you can see I want to work the details better this time. Stay tuned.

I know everyone hates this feature when the controls push the movie up in fullscreen… But what about the playlist? Is it better to push the movie or to overlap it? What is the point of not pushing it if it hides half the movie anyway? …Similar things cross mind.

What about the default design? Should we try to be fancy and put glassy buttons everywhere or stick to rulesPlayer style and keep it simple? Me personally - I like the shiny vista buttons, but I don’t want to see them more than once a day. I prefer vimeo style controls. Really simple, you never get tired of them. That’s why my desktop background is black. You never need to change it, just because there is nothing to change :)

So share your opinion. Make your feature requests and suggestions. It is easy to implement features in the beginning and not so easy after everything is done. As I said, all comments are welcome.

So where we’re at? The current version of the player is pretty stable. It has (almost) everything I wanted and many people seem to like it. It lacks any major bugs and is stable. What I don’t like about it is that the code is a bit dirty… I was actually learning how to embed MPlayer. Anyway I think the result is pretty useful. But there are some other major problems. It’s with my toolkit. First the user interface library that I was using is now dead. Its maintainer went away and I am not keen on maintaining a whole GUI library because of this project. Then the standard library of the language that I am using became not so popular (I have explained in more detail in the forum). To say it plain words, the basic stuff that my player is built on is not useful anymore for my case. And this is a pretty discouraging situation. That kept me from developing the player. Recently I have found a great and very fast GUI library that I used for another project. It is called HTMLayout and it is based on HTML. Which means it offers much greater possibilities for visual presentation. Read skins! This is little cue for the next version - skins. And this where we’re at - next version. This means a complete rewrite. Changing the UI library is a big deal and it means everything will be rewritten. And since I will rewrite everything I have the opportunity to revisit everything that needs a touch and do it the right way. This is risky. Making such a big change there is the risk of breaking what was done right. So it is important to have some kind of guideline. What is most important and what shapes the player as it is? Why the project started in the first place? To have what the others players lack:

• Blazing performance. Both startup and playing the movie (this means fast rewinding). The next version must be even faster.
• Simple, non-bloated interface. The interface must be intuitive to navigate and provide fast access to everything one needs.
• Customizable. One must be able to customize the player. The current player is very customizable, but it is not done the right way. The console approach is not too user friendly.
• Never worry about codecs - they are all included.
• And of course - free.

This is what I intended. Now it must be perfected and extended:

• Eye candy. I am not a big lover of skins but they could be nice if done right. Also there is always the possibility of making Windows like skin.
• More extensible - plugins. It is impossible for one man to please everyone. So I will do what I can - keep the above points. Build a basic player, but make it extensible. So everyone could add what he likes.
• Easier to install and update.

That’s it for now. This is my idea. It may or may not work out. Now is the time for suggestions, feature requests, etc. Comments are welcome. Next time I will give some information about the next version of the player. If you are interested drop by from time to time to see what’s happening.