The Sad State Of FSF's High Priority Projects

[droidhacker]

There is already Marble and there has already been data given by the US and other countries and states to Open Street Map.

TomTom navigation, for example, already allows people to update maps by tracking in case of when road work is being done or streets have changed. It actually works better than the official maps, because those are drawn.

GPS tracking is more precise. It's like MS Office where it gets more usefull the more it is used.

You're missing the entire point, or have no idea what google earth actually IS.
Google earth is about the SATELLITE PICTURES. Not the street maps. I'm not arguing against the utility of GPS traces, but about the INAPPLICABILITY TO THE PROBLEM.

Its like your boss coming to you and telling you that they need rocket guidance software, and you give them a word processor. Your word processor may be really great, but does nothing to guide rockets -- you missed the point.

[V!NCENT]

They want to replace google earth with open source software running with openstreetmap data? How exactly are you going to get satellite imagery from GPS traces? Right, didn't think it would work..... the data HAS to come from *somewhere*, and GPS traces won't provide it.

There is already Marble and there has already been data given by the US and other countries and states to Open Street Map.

TomTom navigation, for example, already allows people to update maps by tracking in case of when road work is being done or streets have changed. It actually works better than the official maps, because those are drawn.

GPS tracking is more precise. It's like MS Office where it gets more usefull the more it is used.

[droidhacker]

They want to replace google earth with open source software running with openstreetmap data? How exactly are you going to get satellite imagery from GPS traces? Right, didn't think it would work..... the data HAS to come from *somewhere*, and GPS traces won't provide it.

[V!NCENT]

Revolution is cool and stuff but first and foremost you have to have something that WORKS and be able to deliver. Linux works, Haiku is in Alpha. HURD is NOT ready.

And that's why releasing Hurd like it is now in a working state is absolutely pointless, because if we want something that works; there is Linux with a lot more drivers already.

So the only way for Hurd to succeed is being at least a thousand times more awesome in terms of design than Linux, otherwise it's just wasted time.

[curaga]

This thread is now about evaluating reindeer meat.

I find it's good.

[kraftman]

Because Linux is boring, old skool and sounds like some local Finnish dish made out reindeer intestines or some other mystery meat, i.e., something you'd really not want to eat.

Indeed! That must be the case. I wonder if the same thinking didn't bring me to Linux years ago?

[DMJC]

Etoile...

Etoile is decent but they seem to be reinventing the wheel a bit. Last time I went into their codebase they had basically cloned System preferences.app under the name of hardware.app or something similar and the individual preferences icons were implemented the same as in system preferences.app in GNUStep. They should have just been making add ons for systempreferences.app. Where they were intending to go originally seemed to be pretty interesting but they appear to have lost their focus recently. Every time I speak to the GNUStep/Etoile developers they seem to have these crazy outlandish ideas of what they want to accomplish with their project (dynamic GUIs building themselves at runtime, you can have a button ANYWHERE!!!! that's nice but I want it in a place I'm used to, not anywhere.), but they never get the implementation down to drive interest. They need to focus on the browser/desktop environment. It has mail/music/irc/text/programming. Web browser/Video/hardware config would finish the base DE so you could use it every day as your only platform. That would drive developers towards programming for it. Right now a gnu step app looks nothing like a gnome/kde app and you need gnome/kde apps to plug the holes in the platform. If they remove those holes, more people will move to the platform. Would anyone use a mac if it had no web browser, no video playback and no display/sound controls? Of course not. GNUStep has the same problem, the developers are trying to run before they can walk.

[joshuapurcell]

You mean like fuse allows on Linux? That is the thing I don't understand about all this "microkernel" stuff. Linux already has support for userspace drivers of several types. If someone really wants userspace drivers, why can't they just code better support for that in the Linux kernel like people have already done for some systems? So rather than just dumping a well-supported and heavily-tested kernel for an entire new one, you slowly migrate (where appropriate) to a more userspace-level design. This seems to be a much more efficient approach.

You mean like the various fuse bindings?

While fuse allows for a user-space file system implementation, you still have the same underlying file system requiring a privileged mode in the kernel. End users get to see some of the benefits of having a user-space file system while still being potentially affected by the underlying architecture's limitations. Here is some more information on the differences between monolithic kernels and microkernels: http://kilobug.free.fr/hurd/pres-en/...tml/node2.html

A system-wide user-space file system is just one benefit the Hurd has over Linux, and there are a few others that I've read about but am not very technically familiar with. The link above points to a good document on the subject. The last thing I would add is that we constantly hear of developers and different types of users wanting to add/remove/change various aspects of the Linux kernel so that it is more suitable for their use. In particular, there is a constant struggle between the server, desktop, laptop, mobile and corporate worlds in regards to the changes they want (and don't want) to see in the kernel. A microkernel would be able to handle these differences much easier for a couple of reasons. The first is that, because the microkernel itself (gnumach for now) is responsible for much less functionality, it would be easier to handle the issues/requests that arise between these competing groups. The other reason is that the Hurd servers can be individually used or not depending on the type of use desired, and completely new Hurd servers which are specific to certain types of uses can be brought up as needed for these otherwise competing users. The Hurd is much more modular than Linux by design.

[XorEaxEax]

Exactly. I don't know what drives some people to look at crap like hurd or haiku when there's Linux.

Well, I'm running Linux as my daily desktop OS and I like it. However, for me, Haiku (and Beos before it) is the best desktop OS experience I've had bar none and it's what I want to use as my day-to-day OS, sadly it's far from mature enough (and perhaps never will be). That doesn't mean that I feel inclined to second-guess other people's choice of OS. You want to use Linux and think it's the best option for you, I see no reason to question your conclusion, same goes for people running Windows, OSX, BSD, etc.

[XorEaxEax]

Not sure that people may want to kill gcc, but replace it probably.

Then why attack GCC? Better to report bugs to LLVM/Clang developers so as to make it better faster. Prefering one product doesn't mean you have to HATE the other. Sadly that is something which seems lost on so many people these days, just look at this board, Linux <> BSD, Gnome <> KDE, GCC <> LLVM/Clang, GPL <> BSD/MIT, etc.

The problem with GCC is until recently everything was in C, the code was almost not understandable and I think it's more related to old coding practices rather than true complex situations.

Definately LLVM's codebase is alot cleaner given that compared to GCC it's quite recent. But 'almost not understandable' comes off as a huge exaggeration given that if it was then GCC development would have died long time ago. Obviously the codebase has been substantially improved over time long before C++ was allowed into the tree. I certainly believe that the competition/existance of LLVM was a big influence on GCC adopting C++ which again underlines that 'competition is GREAT'.

And, gcc was not extensible by plugins for political reasons. So in the end you have a free software (GPL), of course you are free to read the code and modify it, but in fact you can't understand the code and can't replace it because it's too complicated... and limited (no plugins) because of political reasons (GPL).

It is a kind of funny way to loose your "ability" for your "freedom", isn't it ? :-)

Well the debate surrounding plugins and GCC was how it would allow to sidestep GPL and use GCC as frontend/backend for proprietary plugins (for instance Steve Jobs tried to use GCC as a frontend to a proprietary backend for ObjC, and when he wasn't legally able to GCC suddenly got ObjC support through Next). Afaik the current GCC plugin architecture makes it difficult to maintain proprietary plugins which sounds like a decent compromise and as such a carrot to keep them open source, much like drivers in the Linux kernel vs maintaining binary blobs.