what we all need is a new web browser

I think you're right. In the last few days I've tried a few open source browsers for Windows which are considered small, and all of them easily take over 16MB when running. I was surprised by how much memory Firefox takes, although I guess I shouldn't have been, since system requirement for it is 64MB (as opposed to 32MB for some other browsers).

Edit: To be more exact, I tested several browsers, not only open source -- such as Opera. Obviously Opera does have versions which run fine in limited memory, but the latest desktop version takes over 20MB just starting -- but of course, it's quite fancy. I tried a few open source browsers, and only the Gecko based ones seem worth it in terms of what they can display. There's another web rendering engine which is considered quite good (don't remember the name offhand -- it's the one the Mac browser Safari uses), but it'd have to be converted to Windows.

Of the Gecko browsers, K-Meleon, which is supposed to be quite small, starts at over 13MB for the most basic version -- that is, with a very primitive user interface (no icons, no tabs, ...). Note that this is for version 0.8.2. The latest beta doesn't allow choosing what to install, and so it's impossible to compare (it takes 16MB).

I think we need to manage expectations, too... If we can start off small (say, dillo-sized), there is potential that any improvements that people want/need can be put into the base and then inherited into the HPC implementation.

I wonder if Rainer Keuchel could be persuaded to get involved? His current experience in porting Linux/GPL software would be literally invaluable!

Has anyone seen him around?
Cheers,
Nick.

The largest part of the installation for any browser is going to be the HTML interpreter module. If I recall correctly MSHTML on PIE5 is close to 5MB.

Why does it work though. Easy. It's a ROM component.
By creating a new engine you loose the luxury of magically willing away the footprint for the compiler.

So, I think that any port / new HTML object decoder needs to be very rigid. IF it can keep standards support to the here and now, and ignore preparatory and legacy bloat, a new HTML decoder could be made reasonable... I hope.
The UI and the rest can just bolt on as an afterthought. Just like with ftx.

Dillo's footprint is about 500KB of source code(!), and compiles down to around 400KB... I'm thinking that's where we've got to be looking to start from.

We may be able to do something clever to inherit the client-side scripting stuff from PIE. Or we might not. But there's a JSCRIPT/VBSCRIPT engine that is accessible programatically on the platform...

Only under HPC2000. Move down to Pro and it's JS. Move down again to 2.0 and it's neither.
Makes life difficult.

I've kicked off discussions with a Minimo (Mozilla project) leader too.

There is a *LOT* of work to do here. A hell of a lot. However, there is real interest in seeing the project come to light...

So please take this as a call to arms: we need developers. Anyone interested?

Dillo was one of the browsers I looked at when I did my initial search, and I thought about it as a decent candidate for porting (considering that there's an up to date TCL port to CE 2.11). However, after installing it I dismissed it as "not better than PIE". Which doesn't mean that it won't become better with time, but for now it doesn't seem like a real alternative. Gecko and KHTML (the Safari engine) are the only two open source rendering engines I found which seem to be comprehensive enough to provide a basis for an up to date browser. Porting something which doesn't have a better standards support than IE seems like somewhat of a waste of time to me (not a complete one, since it could still be better in some ways, like tabbed browsing, but still not what I'd really look for in a new browser).

BTW, it'd be nice if Opera could be ported, since it's already been ported to other PDA platforms, but I imagine that this will require convincing them that the project could succeed.

Not yet, as far as I'm concerned. I do plan to try my hand at developing for my Handheld PC Pro, but I first have to be able to achieve a working PC connection (this will hopefully happen in a week or two, once I get the CF LAN card I bought on eBay), and then I plan to play with some programming before I even decide whether to try a large project.