Should we port Dillo to Windows CE?

Dillo looks like an excellent web browser that could fix our long-lasting web browsing problem. It is small (source under 400k!!!), written completely in C, and has already been ported to the Jornada 720, but not in Windows CE. AND it is under active development.-If we don't do it, I'm gonna do it myself!!! (Congrats Chiark to pointing out a web browser that is as ridiculously small as this one-just wish that I checked it out earlier)

I'm not too enthusiastic about this, because the Dillo todo list still mentions the need to support frames, there's no internationalisation yet (though it's planned) and SSL is in prototype state. Basically, Dillo is in worse state than PIE 3.0 in some respects, in terms of website compatibility, so I don't see a big gain from porting it. The advantage of the Gecko based browsers, even if they're much bigger, is that the rendering engine is much better featured than Dillo's or PIE's.

Porting Dillo isn't inherently bad, so I won't object loudly if someone does attempt it. I just think that it's not there yet in terms of being a useful alternative.

Edited by ET3D 2004-11-28 6:24 AM

Good point - yet it won't be long until it reaches version 1.0. Maybe we should wait until then, but at the moment it is still more stable than IE. (I just don't like clicking on a link and it doesn't connect until the 3rd try in IE - so I'd like anything but it)

I think porting would be a great idea, any more news on this? I can help out with Beta testing if you like

I gotta finish learning c++/c first... So far I have spent about an hour learning it, and I am learning it very quickly (learning how to make a linux config file first helps). But what I would like is the possibility of other programmers helping me out, though thinking about it I might wait for version 1.0 to come out first, which should be soon. I am looking to fix the web browsing problem very quickly.

Realistically, we need a new browser.

Rather than wander off in different directions, I think we need to determine requirements, and whether Dillo fits them.

I suggest downloading Dillo and running it as your main browser on your PC for a few weeks to find its strengths and limitations. I suspect that rapidly you'll want more functionality. It *is* lightweight, but it *is* limited currently.

I don't know if you've looked at the code, but I'm guessing that the statement "I'm gonna do it" means that you've reviewed it extensively and know how to do it.

Dillo is coded in pure C, not C++, uses GLIB and GTK, and would be best compiled using the GNU Toolset - so look at mamich's port of GCC etc. The pure C approach means that the author has come up with inventive ways to replicate some C++-style functionality within the framework he's created, so you'd need to understand the coding style and usage of various mechanisms. This is IMHO non-trivial.

I have tried finding a port of GTK for WinCE, which would make the task a hell of a lot easier, but it doesn't exist. There is a port for Win32, but again, check the size of the task you're undertaking - porting GTK to WinCE is IMHO not worth it for DILLO, as you'd be porting megabytes of tookset for a 400KB application. On the other hand, having GTK on WinCE could open up a lot of new possibilities, such as the GIMP?

The GTK stuff could conceivably be handled using a stub API that you write. I would not want to try to replace the GLIB stuff, as that will be messy.

I mailed the DILLO developer a few weeks back and got a response immediately. He would definitely like to see a CE port...

Personally, I think a smaller starting point such as Lynx or Links would be a better bet, as it would expose the team to the Joys of Web Client Programming, and avoid much of the UI which will be the least of our worries!

I still think we need to round up developers and work as a group on this, as even Dillo (400KN source) is non trivial to port for the reasons above.

If you've not written any C or C++ on any platform, let alone WinCE, I honestly suggest you start a little smaller. Pick a small but useful app on SourceForge (or wherever) and set about porting that. It'll show you some of the things you need to consider...

Cheers,
Nick.

I will progress, just give me time. Yes, I had thought about links earlier, and that would probably be very easy to do, as it runs off the command line (I mean the text-only version) it most likely isn't built off of a toolkit, or at least the older versions. (That should be checked into though) I thought that there was already a way to get that to work on Windows CE though...but anyways it looks like a lot of stuff is going to start being ported to this platform very soon. I'll work on this later as at the moment I am really stalled as far as working on learning anything, until a few weeks from now.

Hey lookie at this: http://www.akos.uklinux.net/tclce/ TCL 8.3 on CE! (This is probably useless for now, but could probably come in handy)

TCL/TK 8.4.6 is available at http://sourceforge.net/projects/tcltkce, although I'm not sure which CE versions it supports, since I couldn't find any documentation. There are ARM and MIPS downloads. TCL will be useful for porting BrowseX (http://browsex.com/) -- another small browser I don't consider very port worthy (although it may be an interesting alternative to links as a simple option, assuming that the TCL part is easy to get working in CE).

Yeah browsex does have the classic tcl graphics and would probable be easier to port. But both Dillo and Browsex are limited. I'll check browsex out a little more to see if it is port-worthy. Has anyone sent an email to the firefox team to find out if they would like to see it ported to CE? (I think we might want to stick with the light ones, stated above, though I may find more...

-Anyways, we really should decide what we want to do. It looks like whatever we decide to do will need more things to be ported. Though, this could avoid the problems of having to create a browser ourselves. Pretty much, all we need is basic functionality, I really can't imagine running java on my hpc. I will check to see all that these browsers can do, they might have plugins. But pretty much I think something similar to ftxbrowser running on a different engine would be better. Maybe we could find out if the core gecko engine is small enough, but I don't know...

Edited by programsynthesiser 2004-11-30 8:06 PM

Ugh never mind about dillo. I was just writing an ok...kinda review on it, then I realized that I couldn't log in or submit my review on this forum. The submit button fix for it didn't work either. I guess if we want to port some browser then we should check into something console-related, maybe something that we could make run in the CE console. I'm leaning toward w3m, links, lynx etc. here. It would be nice if we could just get some sort of alternative browser running, as I actually like browsing in text.

I have emailed the Minimo group of Firefox, and they're interested in seeing a WinCE port - in fact, there has been some thought invested in such a thing already by one of the project leaders.

Check out Rainer Keuchel's webpages for information on stuff that has already been ported to WinCE, and also for a very handy CE Library which will definitely assist with any porting that you're doing. Rainer has tried to compile Minimo, but has so far struck a blank even with the compilation on a Win32 machine. Last I heard from him, he was asking me some questions about the structure and I offered to put him in touch with the Minimo team - I haven't heard back.

Drop me a mail if you like, and I'll pass on the contacts, but I think you'd be better doing this once you've got a bit of experience under your belt. Please don't take that as a lack of faith in your abilities, I just think you need to be exposed to the development environment and language, that's all

I still think we (the HPC Community) are better doing this as a team effort rather than individually. No matter how motivated you are, porting a browser of the complexity of Firefox (for example) would be a full time job for months and months for an experienced C++ / Win32 / WinCE developer.

Good find on the TCL toolkit - that might make some things easier. However, I think the problem you're going to face is more in the internals of a web browser rather than the UI - get the internals right, and the UI can come later (providing you've got enough of a stub to input/output to the engine)

Perhaps we should resurrect the "call to arms" again now we've got a few more on the forums?

Cheers,
Nick.

Yeah, I was thinking about the call to arms to. All further posts should be there as this is going away from Dillo now. Talking about Rainer, I thought he stopped porting, as it seems that it has been a long time since he has any development, e.g. his ssh shell has been at alpha I think since 2001? (Why isn't he on this forum? ) Yes, it will be a while until I get my programming skills up. I have spent so much time learning about operating systems that I never bothered to learn to code (Though if we ever do port programs from other OS's, that will come i very handy). So is there already a person working on porting Minimo other than Rainer? (post reply on call to arms I guess)

I'll reply here as we're almost still on topic

The minimo leader has invested some time in considering a port to WinCE, and I've had a few emails with him. That's the extent of work on the Minimo -> WinCE port.

Rainer is quite elusive, but I was glad to get a reply from him, and for him still to be interested. I'll pick up that thread of conversation with him again.

But, in short, unless we can find people to work on the project, it isn't going to happen

Cheers,
Nick.

Grr you just can't learn something fast enough. I haven't had time to go through the tutorial lately. Maybe we should postpone this before we get screwed. We'll need a bit more time to get developers together before starting the project. This is starting to sound a bit dangerous here (in programming/starting means ) C:Amie, where is that whip?! lol