"They should shutter their business and release all their developers to go work on real projects that have a hope of succeeding."
While their developers are probably first rate, they have been trained by the M$ management to produce sub-stellar products. If M$ goes away, all of those managers would infect other companies. Can the world really afford that?
True. That's just what happened when Compaq went down.
And that is exactly what Google is trying to fix by the project. Did you see OP?
Of course. And I saw plenty of high-sounding initiatives at Google that had no useful result. I am skeptical about this one. It plays well, but does Laszlo really have his heart in it or is he just going through the motions, hoping that whatever concern came down from the owners will dry up and blow away after a while and everything can go back to business as usual?
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't.
It's capable of implementing a better object-oriented system than what C++ provides...
You're smoking something or you have never written any respectable amount of C++. C++ just provides you the mechanism, it's up to you whether you abuse it or leverage it. C simply doesn't provide the necessary mechanism and the nasty hack called GObject does not come close to fixing that.
Sorry about that, your thoughtful comment deserved better. To put it simply, the Google you see from the outside and the Google that actually is are two different things. Even when you visit, you don't really see inside. Google is indeed falling into a number of traps, which you would think that as certified smart people they would recognize and avoid. But that's where the Google myth is already kicking in. You see, Google isn't really full of smart people, it's actually full of entirely typical schmucks like you and me.
Laszlo just demonstrates that by selling the value of technical expertise short, he is part of the problem. But I already knew that. In all fairness, Laszlo is really the reason for the majority of management dysfunctionality at Google because he spent way too many years looking the other way as frontline managers make a mockery of the systems that were put in place. Eric having his head in the clouds didn't help.
Are you serious? It is rare to find one person with good technical skills at an entire company. I worked at a nameless multinational software company for 4 years recently before I moved on to another company and I wouldn't trust most of my fellow engineers at that company to screw in a light bulb.
Who was responsible for employing and retaining these less than apt people?
while buffer objects are great and all for the main game engine, there's something to be said for immediate mode and/or display lists for some of the less performance dependent stuff.
Totally agree. I always start by writing stuff in immediate mode, it is significantly less error prone and easier to debug. But I also completely agree that immediate mode should be a separate profile than core, which should always be your production target. If you want some of your core engine to work like immediate mode, then implement a little shell that does it, this is just a quick toss off. Or use one of the many that already exist. When you are ready to productize you can easily justify this finishing work. Being able to have a really clean core tightly coupled to the underlying hardware is a huge blessing.
Google is known for hiring very smart, very technical people, then abusing and humiliating them. There are exceptions, if you are one of them then count your blessings, but this is the prevailing climate at Google today. I don't know how many truly awe inspiring, highly educated people I saw stuck in crap jobs there doing things like rebooting servers while their managers are off running around the countryside getting drunk at offsites and stroking each other about what smart people they are.
Just like the maturation phase of every other technology focused corporation in history...
1. Founded by engineers 2. Rapid growth 3. Founding engineers become wealthy and retire early 4. Sales, marketing and management folks take over 5. Bureaucratic creativity sucking shithole
You nailed it precisely. So sad, I expected better of Google.
Another thing: managers at Google widely believe that they are better than engineers simply because they are managers, in spite of a supposed explicit ban on this attitude. For that matter, so do the sysops, because they are in control of the facilities engineers need to do their work, and because they get first dibs on any shiny new equipment that arrives. I got the distinct impression that Google sysops think of themselves as managers, or at least, very important people, and in particular, more important than engineers. By the way, I was a Google sysop before I moved to engineering so I saw this from the inside.
And I can tell you that the general quality of Google managers is very poor in spite of supposed systems for filtering, training and guiding them. This is in fact the worst thing about working at Google: self important, self absorbed managers who only care about milking their own situation for everything they can get. Often nonexistent or weak technical skills just pours salt on this bleeding wound.
The few guidelines that Google puts in place tend to be unmonitored by anyone who matters and are widely and cynically ignored. Peer review is very much one of those. There are of course good managers at Google, I know a few. But they are badly outnumbered by facetimers and soulless climbers.
It doesn't. That's why people working remote often go visit the people they are working with, or at least they have one person who does if there are a group of them. Telecommuting works because there is a buffer of understanding built up by in-person meetings and actions.
You are an iTroll who has never telecommuted. Would you please quit spouting about things you know nothing about. I have telecommuted most of my professional career, at least since the internet was invented. The occasional in-person meeting is nice, but not necessary. Video conferencing works fine, IRC or similar work even better because they get the same job done without wasting as much collective time. Conferences and face to face are mostly fun junkets that are expensive time wasters. But fun, and sometimes needed to avoid cabin fever. Otherwise, telecommuting makes the open source world go round. Most of the big names work from home, as is their well earned prerogative.
KDE 3 was fine, so I could still use trinity, but I ended up with LXDE. It does look like Windows 95, but it is light, fast, and not in the way. Perfect for running 4 lxterminals:-)
KDE 4 is now fine as well. I run it on some pretty low spec machines (atom and worse). KDE blows the living heck out of Xfce in terms of usability, features, customization, integration and prettiness. Pretty much the same story vs Gnome to a somewhat lesser extent. I have worked with internals of both Xfce and Gnome, and they are both pretty disgusting ad hoc piles of spagetti. KDE's internal structure, based on QT and slots as it is, is relatively more approachable and capable of implementing complex component interaction with a hope of not leaking resources or getting its internal wires crossed.
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't. What you end up with is an unholy unmaintainable mess. Full of messy casts, and full of bugs
Funnily enough, someone realized that, and they've actually made a programming language (syntactically largely a C# derivative) that is based on the GObject object model, but which hides all of the ugliness that is inherently exposed in C. It's pretty neat, as the output you get is still usable directly from C, but your code is high-level and bug-free (at least in parts which have to do with GObject plumbing).
What a perfectly horrible idea, a not-quite-C#-wanabee to hack around the fact that C is not C++. Ranks right up there with Corba, Bonobo, Gconf and a long string of bad Gnome ideas. Stupid question: if you're going to rewrite, how about rewriting in proven language rather than yet another experimental flight of fancy?
...you never hear GNOME's side of the situation, making a lot of disrespectful comments about GNOME (or the others involved) rather baseless and illogical
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't. What you end up with is an unholy unmaintainable mess. Full of messy casts, and full of bugs, as Gnome has always been. And unable to express reasonable defaults for things in any powerful or consistent way, the result being that Gnome tends to have lousy defaults for just about everything. Add in a liberal does of hubris from certain Gnome maintainers, and have blinkers on regarding the limitations of the GUI toolkit, and you have a recipe for the nasty mess that Gnome has been from the get-go, and will be until somebody finally does something about it. Thankyou Mark.
Incidentally, these comments apply equally to glib, dbus and various other crappy decorations the Gnome guys have forced on the Linux desktop over the years. At least Bonobo is gone, that goodness for small mercies.
Consoles have exactly one thing going for them: convenience. There is no way consoles can ever again compete seriously in the high end gaming arena.
I'm not sure I agree with that. I think if they got cheap enough to buy multiples, and if enough games actually supported multi-console multi-screen gaming, then I think people would do that. In fact the consoles could be offered either without an optical drive or in both optical and non-optical models to make it more affordable; non-optical consoles would still be able to download and play games, so some people would never buy anything else.
Of course, there are major technological hurdles to be jumped, therefore it won't happen any time soon, so I do suspect there will be a lull of the type you describe.
That lull is a death sentence. Cheap and powerful are mutually incompatible because of the laws of physics and economics. Of course by powerful I mean in comparison to the PC (and coming soon, Arm) market. It's just not feasible for next generation consoles to compete at the same performance level as PCs. It was barely possible in the current generation and both Microsoft and Sony suffered embarrassing meltdown issues costing billions of dollars. Ballmer would be quietly assassinated by institutional shareholders before they let that happen again, and Sony was so badly wounded that next time round they either show more sense or die.
Mind, I'm totally happy Sony shot their wad in this generation, it meant I could play games without booting Windows. Next generation is a different story, we don't need Sony any more to get away from Microsoft.
You need to rewrite for a mobile platform anyway. Unless you're using pretty basic functionality in which case you might as well use OpenGL in the first place. Except you don't have display lists or immediate rendering. Or fragment shaders (unless you use 2'0 in which case you don't have a matrix stack).
If you have to rewrite more than the UI for a mobile platform you made serious design errors in the first place. (And if your game mostly consists of UI then nobody should but it.) Nobody uses display lists or immediate mode for a production game engine any more. Everything is arrays and buffer objects now, which are the same on Opengl ES (sorry for mis-writing EGL earlier) and every version of OpenGL from 2.0 on. Everybody uses their own matrix library now. And soon, every Android phone that matters will support shaders. There are well known techniques for writing lovely 3D graphics without shaders, which is where the first generation touch screen gaming market will sit because it has no choice. Such code works perfectly well on PCs as well, particularly low end netbooks and the like.
Athletic would have seen most of OpenGL 3 in OpenGL 2. Khronos is playing catchup.
Absolutely true. Before Kronos the history of OpenGL was all about Microsoft bodychecking SGI headfirst into the corner. The landscape is quite different today, and ironically, that is in part due to Microsoft having taken down SGI.
Slashdot Mirror
"They should shutter their business and release all their developers to go work on real projects that have a hope of succeeding."
While their developers are probably first rate, they have been trained by the M$ management to produce sub-stellar products. If M$ goes away, all of those managers would infect other companies. Can the world really afford that?
True. That's just what happened when Compaq went down.
Part of your viewpoint is that you worked in the SMO office.
True. But I spent time in Mountain View as well and observed similar situations, just not as shamelessly out of control.
And that is exactly what Google is trying to fix by the project. Did you see OP?
Of course. And I saw plenty of high-sounding initiatives at Google that had no useful result. I am skeptical about this one. It plays well, but does Laszlo really have his heart in it or is he just going through the motions, hoping that whatever concern came down from the owners will dry up and blow away after a while and everything can go back to business as usual?
This is just human nature, and it will not change.
Not if you just throw up your hands and accept it. That sort of attitude can be very costly to shareholders.
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't.
It's capable of implementing a better object-oriented system than what C++ provides...
You're smoking something or you have never written any respectable amount of C++. C++ just provides you the mechanism, it's up to you whether you abuse it or leverage it. C simply doesn't provide the necessary mechanism and the nasty hack called GObject does not come close to fixing that.
Yes, objective C would also be viable, maybe better from a Gnome mindset.
So, PHBs and BOFHs
Yup.
Sorry about that, your thoughtful comment deserved better. To put it simply, the Google you see from the outside and the Google that actually is are two different things. Even when you visit, you don't really see inside. Google is indeed falling into a number of traps, which you would think that as certified smart people they would recognize and avoid. But that's where the Google myth is already kicking in. You see, Google isn't really full of smart people, it's actually full of entirely typical schmucks like you and me.
...Admittedly, I've not worked there...
Nuff said.
Laszlo just demonstrates that by selling the value of technical expertise short, he is part of the problem. But I already knew that. In all fairness, Laszlo is really the reason for the majority of management dysfunctionality at Google because he spent way too many years looking the other way as frontline managers make a mockery of the systems that were put in place. Eric having his head in the clouds didn't help.
Are you serious? It is rare to find one person with good technical skills at an entire company. I worked at a nameless multinational software company for 4 years recently before I moved on to another company and I wouldn't trust most of my fellow engineers at that company to screw in a light bulb.
Who was responsible for employing and retaining these less than apt people?
Corporate America?
True to some extent. But try pissing off a sysop or two and see how that works out for you.
while buffer objects are great and all for the main game engine, there's something to be said for immediate mode and/or display lists for some of the less performance dependent stuff.
Totally agree. I always start by writing stuff in immediate mode, it is significantly less error prone and easier to debug. But I also completely agree that immediate mode should be a separate profile than core, which should always be your production target. If you want some of your core engine to work like immediate mode, then implement a little shell that does it, this is just a quick toss off. Or use one of the many that already exist. When you are ready to productize you can easily justify this finishing work. Being able to have a really clean core tightly coupled to the underlying hardware is a huge blessing.
Google is known for hiring very smart, very technical people, then abusing and humiliating them. There are exceptions, if you are one of them then count your blessings, but this is the prevailing climate at Google today. I don't know how many truly awe inspiring, highly educated people I saw stuck in crap jobs there doing things like rebooting servers while their managers are off running around the countryside getting drunk at offsites and stroking each other about what smart people they are.
Just like the maturation phase of every other technology focused corporation in history...
1. Founded by engineers
2. Rapid growth
3. Founding engineers become wealthy and retire early
4. Sales, marketing and management folks take over
5. Bureaucratic creativity sucking shithole
You nailed it precisely. So sad, I expected better of Google.
Another thing: managers at Google widely believe that they are better than engineers simply because they are managers, in spite of a supposed explicit ban on this attitude. For that matter, so do the sysops, because they are in control of the facilities engineers need to do their work, and because they get first dibs on any shiny new equipment that arrives. I got the distinct impression that Google sysops think of themselves as managers, or at least, very important people, and in particular, more important than engineers. By the way, I was a Google sysop before I moved to engineering so I saw this from the inside.
And I can tell you that the general quality of Google managers is very poor in spite of supposed systems for filtering, training and guiding them. This is in fact the worst thing about working at Google: self important, self absorbed managers who only care about milking their own situation for everything they can get. Often nonexistent or weak technical skills just pours salt on this bleeding wound.
The few guidelines that Google puts in place tend to be unmonitored by anyone who matters and are widely and cynically ignored. Peer review is very much one of those. There are of course good managers at Google, I know a few. But they are badly outnumbered by facetimers and soulless climbers.
It doesn't. That's why people working remote often go visit the people they are working with, or at least they have one person who does if there are a group of them. Telecommuting works because there is a buffer of understanding built up by in-person meetings and actions.
You are an iTroll who has never telecommuted. Would you please quit spouting about things you know nothing about. I have telecommuted most of my professional career, at least since the internet was invented. The occasional in-person meeting is nice, but not necessary. Video conferencing works fine, IRC or similar work even better because they get the same job done without wasting as much collective time. Conferences and face to face are mostly fun junkets that are expensive time wasters. But fun, and sometimes needed to avoid cabin fever. Otherwise, telecommuting makes the open source world go round. Most of the big names work from home, as is their well earned prerogative.
KDE 3 was fine, so I could still use trinity, but I ended up with LXDE. It does look like Windows 95, but it is light, fast, and not in the way. Perfect for running 4 lxterminals :-)
KDE 4 is now fine as well. I run it on some pretty low spec machines (atom and worse). KDE blows the living heck out of Xfce in terms of usability, features, customization, integration and prettiness. Pretty much the same story vs Gnome to a somewhat lesser extent. I have worked with internals of both Xfce and Gnome, and they are both pretty disgusting ad hoc piles of spagetti. KDE's internal structure, based on QT and slots as it is, is relatively more approachable and capable of implementing complex component interaction with a hope of not leaking resources or getting its internal wires crossed.
And what proven language would that be, honey? Surely not C++
Surely c++, which rules the world at the moment. To say otherwise would be delusional.
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't. What you end up with is an unholy unmaintainable mess. Full of messy casts, and full of bugs
Funnily enough, someone realized that, and they've actually made a programming language (syntactically largely a C# derivative) that is based on the GObject object model, but which hides all of the ugliness that is inherently exposed in C. It's pretty neat, as the output you get is still usable directly from C, but your code is high-level and bug-free (at least in parts which have to do with GObject plumbing).
What a perfectly horrible idea, a not-quite-C#-wanabee to hack around the fact that C is not C++. Ranks right up there with Corba, Bonobo, Gconf and a long string of bad Gnome ideas. Stupid question: if you're going to rewrite, how about rewriting in proven language rather than yet another experimental flight of fancy?
...you never hear GNOME's side of the situation, making a lot of disrespectful comments about GNOME (or the others involved) rather baseless and illogical
Some negative comments about Gnome are not at all baseless, for example the one I am about to make. Gnome is based on an outmoded hack of an attempt to build an OOP based GUI without the benefit of an object-oriented compiler. Instead is uses a collection of nasty hacks and conventions, which which I am deeply familiar because I once was deluded enough to think also that C is just as capable of writing object object oriented code as C++. It isn't. What you end up with is an unholy unmaintainable mess. Full of messy casts, and full of bugs, as Gnome has always been. And unable to express reasonable defaults for things in any powerful or consistent way, the result being that Gnome tends to have lousy defaults for just about everything. Add in a liberal does of hubris from certain Gnome maintainers, and have blinkers on regarding the limitations of the GUI toolkit, and you have a recipe for the nasty mess that Gnome has been from the get-go, and will be until somebody finally does something about it. Thankyou Mark.
Incidentally, these comments apply equally to glib, dbus and various other crappy decorations the Gnome guys have forced on the Linux desktop over the years. At least Bonobo is gone, that goodness for small mercies.
Consoles have exactly one thing going for them: convenience. There is no way consoles can ever again compete seriously in the high end gaming arena.
I'm not sure I agree with that. I think if they got cheap enough to buy multiples, and if enough games actually supported multi-console multi-screen gaming, then I think people would do that. In fact the consoles could be offered either without an optical drive or in both optical and non-optical models to make it more affordable; non-optical consoles would still be able to download and play games, so some people would never buy anything else.
Of course, there are major technological hurdles to be jumped, therefore it won't happen any time soon, so I do suspect there will be a lull of the type you describe.
That lull is a death sentence. Cheap and powerful are mutually incompatible because of the laws of physics and economics. Of course by powerful I mean in comparison to the PC (and coming soon, Arm) market. It's just not feasible for next generation consoles to compete at the same performance level as PCs. It was barely possible in the current generation and both Microsoft and Sony suffered embarrassing meltdown issues costing billions of dollars. Ballmer would be quietly assassinated by institutional shareholders before they let that happen again, and Sony was so badly wounded that next time round they either show more sense or die.
Mind, I'm totally happy Sony shot their wad in this generation, it meant I could play games without booting Windows. Next generation is a different story, we don't need Sony any more to get away from Microsoft.
You need to rewrite for a mobile platform anyway. Unless you're using pretty basic functionality in which case you might as well use OpenGL in the first place. Except you don't have display lists or immediate rendering. Or fragment shaders (unless you use 2'0 in which case you don't have a matrix stack).
If you have to rewrite more than the UI for a mobile platform you made serious design errors in the first place. (And if your game mostly consists of UI then nobody should but it.) Nobody uses display lists or immediate mode for a production game engine any more. Everything is arrays and buffer objects now, which are the same on Opengl ES (sorry for mis-writing EGL earlier) and every version of OpenGL from 2.0 on. Everybody uses their own matrix library now. And soon, every Android phone that matters will support shaders. There are well known techniques for writing lovely 3D graphics without shaders, which is where the first generation touch screen gaming market will sit because it has no choice. Such code works perfectly well on PCs as well, particularly low end netbooks and the like.
Athletic would have seen most of OpenGL 3 in OpenGL 2. Khronos is playing catchup.
Absolutely true. Before Kronos the history of OpenGL was all about Microsoft bodychecking SGI headfirst into the corner. The landscape is quite different today, and ironically, that is in part due to Microsoft having taken down SGI.