Evolved Virtual Creatures

A pretty cool video showing evolved virtual creatures:

More infos here, with Karl Sims Siggraph's 1994 paper..

another video from Nicolas Lassabe

more on Lee Graham's page :

Cairo backend

Just a quick note: Fred Kiefer fixed the last issue I had (bad text clipping / smearing) with the Cairo backend... so, no more little glitches, it's really usable now !

Dock

Looks like it didn't take long for Yen-ju to play with the Cairo backend and partial transparency :)

Shiny. (yes, too much like OS X)

GNUstep Cairo backend

After some effort, I finally managed to find a solution to the annoying scrolling bug in the Cairo backend... There's still some little graphic glitches appearing sometimes -- it looks like some clipping/refresh issues -- but overall it's nearly there :)

For the intrepid, the code is committed on the gnustep trunk. The above screenshot shows the Etoile desktop (with the new compositing manager) and the work-in-progress Narcissus theme, running with the Cairo backend.

AlpenStep '07

I was at AlpenStep this weekend, you can check some pictures.

It was a very cool event, thanks in particular to gerold's organisation. The location, in a small swiss village, was really nice -- I can say that some fresh air after beeing in london is good !

It was also a small-scale event -- only 7 persons (though gurkan came but left saturday before lunch :-/), which made it a very intensive event as well: lots of discussion, lots of coding too. Fred Kiefer was probably the busiest person in the room, as everybody wanted to ask him questions/advices... ;-)

Riccardo Mottola gave a nice GAP demonstration, and during the weekend he and Nikolaus Schaller were busy working on SimpleWebKit:

Nikolaus also made a cool presentation of QuantumSTEP running on the OpenMoko platform :)

The OpenMoko device is actually fairly nice, very high-res screen, and some interesting hardware to tinker with.

Quentin was working on ContainerKit for Etoile, a very interesting way of specifying UI using a generic data model (so you can switch representations on the fly for instance), also allowing very nice introspection features with inspectors generated on the fly. I tried to convaince him to introduce some Magritte-like magic, i.e. adding metadatas to your model...

Quentin and I also made a presention of Etoile, describing the various components/frameworks/applications, and how everything is supposed to fit...

Finally, I can only be admirative of gerold organisation skills... to the point were he got us a GNUstep cake:

Which was delicious !

In short, looking forward for AlpenStep'08 ! (and Fosdem'08 before...)

Cairo/GNUstep

I was at AlpenStep this weekend (photos will follow..) which was quite cool. I teamed with Fred Kiefer (the GNUstep gui maintainer) to try to iron out some of the quirks of the Cairo backend. The main problem (recopy of surfaces is not always done well, which impact scrolling) is still here, although we improved some things, and it looks now like it should be solvable with a bit more work. We also added 32 bit surface support when choosing x11 visuals : a cool side-effect is that you can then use directly the alpha channel on the window :) and thus draw semi-transparent windows.

David will like that ;-) -- he's working at the moment on Etoile's compositing manager...

In the above screenshot, I simply fill the view with NSCompositeClear (so it's just transparent), and then composite an image on it, with 0.8 transparency. You can then see that parts of the window are fully solid, while some are completely transparent, and others partially transparent.

Et merde.

On est pas sorti de l'auberge maintenant. Ca donne pas envie de revenir en france en tout cas... :-/


Resultats de l'election presidentielle

Hard Disk Crash

...You know when people tell you to do backups ? Feeling guilty ? :-)

Well, most of my data is usually backed up, but... even so, there's still a lag usually, particularly for "not-so-important data", etc. Add to this usual mix some data put away "temporarily" from my main computer... and that's when my LaCie 250Gb external drive choose to die. Around 200Gb of data lost, among them some that I could get back with various difficulty somewhere else, eg on some dvd or other computers (the LaCie *was* the main/more used backup plan), and some truly lost (yes, this "temporarily moved" data, yes, it's you I'm talking about), or some data recently copied only on the LaCie... Of course this drive is around 3 years old, and I actually had some thoughts recently of doing a proper backup of everything there was on it, for convenience, and for fear of such an accident. Of course too, I ended up way too busy to actually do it...

Still, a bit of luck: there was no scratching noise, in fact, there wasn't any noise at all: the disk just didn't start, which meant that it was more likely an electronic fault than a mechanical one. Removing the hdd from the LaCie enclosure and plugging it directly on a computer didn't do anything either (so it sadly wasn't the LaCie controller board). So that left us with the actual hdd electronic board... I then looked on the web for an hdd from the same brand and the same model -- and found a cheap one from an online reseller. After receiving it, I checked it -- yes, it was the same model, yes, the same year, yes the same firmware... ah. No. not the same firmware ! well, still, only two months separated those drives, so I crossed my fingers and started removing the controlling board of the "new" drive to swap it with the one on my crashed drive. Never having done that, I wasn't quite sure what to expect :-) but it's in fact really straightforward (you just need the right kind of screwdrivers -- Torx). The only difficulty is in finding a control board that is compatible with your hdd!

To cut the story short, the "patched" drive started happily and worked fine, so I passed the day swapping and burning dvd of the drive's content ;-)

One thing for sure is that I will seriously think of building a proper backup solution (ie, automated/redundant) for hosting "home" data (photos, videos, music...) asap... I was lucky to save this drive ! these kind of content is annoying as they quickly take so much space, yet (for at least some of them) they are invaluable...

NeXT videos & WebObjects

I found these two videos, the first one showing Steve Jobs demo NeXTstep 2.0, the NeXTCube 040 + Dimension, the slab, DBKit, in 1990.. it's always surprising to see how far ahead NeXT was, or reversely, how slowly did we progress on the software side, while the hardware we have now is insanely more powerful...

The second one shows again Steve Jobs, five years later. NeXT stations didn't sell, their state-of-the-art factory closed, they stopped producing hardware, and reconverted as a software only company. The tone of the talk seems to reflect those disappointment, with Steve trying to be excited about Distributed OLE and Portable Distributed Object (PDO), but you can feel it's not really anymore about changing the world... and yet, when analysing the state of the WWW and its future, Steve is absolutely dead on, and he then goes on to demo a beta version of WebObjects, which in the end will save the company (until it took over Apple). WebObjects was an absolutely amazing tech, and still is. The move to a J2EE implementation instead of the ObjC implementation, the availability of other solutions, and its curious downplay by Apple (while they use it for their store, they aren't exactly pushing it on the market, although they dramatically cut the price from hundred thousands of dollars to $500, to finally make it free). WebObjects had a huge impact on current "web application servers", many of them openly inspired by it. My preferred one is Seaside, a fantastic Smalltalk framework... for a great example of what you can do with Seaside, try dabbledb :-)

Note that there's two free software implementations of WebObjects in Objective-C, GNUstepWeb and NGObjWeb (part of SOPE, the web application framework used by Opengroupware.org).

GNUstep participates in Google Summer of Code 2007

The GNUstep project just announced that it was accepted in the Google Summer of Code 2007 :

GNUstep provides a cross-platform solution for Objective-C/OpenStep/Cocoa developers. This year, it offers various projects for SoC, such as adding new classes from Mac OSX 10.4, enhancing text system, porting WebKit to GNUstep, improve GNUstep on MS Windows platform, etc. It is a great chance for students to learn programming in many aspects.



read more | digg story

It's an excellent news ! we tried to be accepted first two years ago if I remember, and last year we were in but under the GNU umbrella... and no projects were accepted. Beeing accepted as a separate organisation should help us, hopefully !

Beside, any progresses on GNUstep will be immediately useful for Étoilé :-)

If you are a student, it's definitely worth a look -- GNUstep is one of the best OO api available here, and a fun place to hack.

update: Squeak is also in ! :-) great news too...

Pink Floyd

Today I was in london for a job interview, and had a few hours to kill after it before going back to wales.. so I walked around a bit, and ended up reaching the Thames (duh! not surprising in london, is it ?). But I just ended up in front of the Battersea Power Station :

Of course, this power station is world famous because of Animals, the 1977 Pink Floyd album... ;-)

Sadly, I didn't have my camera with me, so the shots are from my mobile... here's some more:

ps: in the "cool for a geek dpt", I'm actually posting that from my laptop.. but through my 3G mobile cnx. With a cheap unlimited data plan, you basically get internet access everywhere -- as like here, in the train :D

Fun with Objective-C

Following a mail from david on étoilé dev about how it could be nice to try allocating ObjC objects on the stack, I played with a couple of things...

First I wrote a mini benchmark -- get a very basic object (a couple of ivars, one method "plop" assigning a value to an ivar), and then create an instance, initialize it (call the init method), call the one method the object has, then deallocate the object. 10000000 times. As far as micro benchark goes this one is pretty stupid but well, that'll give us some ideas of what's going on.

Without optimizations: ~4.10s (on a macbook pro 2.16Ghz, core duo)

Not that great -- doing the same thing in C++ with a similar object here is the timings I get:

Objects created on the stack: 0.24s (17 times faster !)
Objects created on the heap: 1.37s (3 times faster)

Ouch, the poor Objective-C... not a surprise when allocating objects on the stack, but even allocating them on the heap C++ is still 3 times faster.

One obvious reason is that Objective-C calls a lot of methods when creating an object; and a method call is more costly than a C++ function call. Still...

So the obvious idea here is to cache some method calls (ask their address then call the function directly -- as a C function). I restrained myself to alloc/init, the method the object had ("plop"), and the release method. Of course, there's other methods that are called by those, that won't be cached.

Caching method calls: 2.77s

It's still twice as slow as creating the C++ object on the heap, but it's anyway a nice performance increase.

Ok, then our only option left is to allocate the Objective-C object on the stack. Surprise:

ObjC objects created on the stack: 0.23s
ObjC objects created on the stack + cached imps: 0.13s

:-)

[note of course that it's a stupid micro-benchmark that doesn't prove much, but it's fun]


...


What ? how can you allocate objective-c objects on the stack ? ah well... you can:

#define STACKCLASS(class) typedef struct { @defs(class) } \
    __CLASS_ON_STACK__ ## class;


#define STACKOBJECTISA(objectName,className,classIsa) \
    __CLASS_ON_STACK__ ## className __INSTANCE_ON_STACK__ ## objectName; \
    __INSTANCE_ON_STACK__ ## objectName.isa = classIsa; \
    className* objectName = (className*)& __INSTANCE_ON_STACK__ ## objectName;


#define STACKOBJECT(objectName,className) \
    STACKOBJECTISA(objectName,className,[className class]);

Here is an example:

STACKCLASS(Test);

int i;
for (i=0; i< 10000000; i++)
{
  STACKOBJECT(test,Test);
  [test init];
  [test plop];
}

Basically, create the corresponding struct for the class and set the isa member (you can also cache the class isa to gain one message send). A bit of a hack, but that seem to work ok :)

Note of course that by doing that, you loose flexibility -- exit class clusters for instance (eg classes that actually returns another class instance, such as NSNumber), you can only work with concrete classes. And also, don't call directly -dealloc as it would try to deallocate the object, which is not needed as it was created on the stack (so if you need to do some cleanup you should do it in another method).

Here's the macros I used for imp caching (fairly straightforward):

#define CALLIMP(imp,object,sel,args...) \
    (*imp)(object, @selector(sel) , ##args)
#define GETIMP(class,sel) [class methodForSelector: @selector(sel)];

You use them like that:

IMP imp1 = GETIMP(Test,alloc);
id p = [Test new];
IMP imp2 = GETIMP(p,init);
IMP imp3 = GETIMP(p,plop);
IMP imp4 = GETIMP(p,release);
[p release];

int i;
for (i=0; i< 10000000; i++)
{
   id test = CALLIMP (imp1, c, alloc);
   CALLIMP (imp2, test, init);
   CALLIMP (imp4, test, plop);
   CALLIMP (imp3, test, release);
}