Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999
Wishiknew writes, "In a 1 1/2 hour lecture to a University of Washington's computer science class, Chief Scientist Bill Joy of Sun Microsystems Inc. criticized Netscape Communications 'Mozilla' effort for it's inability to 'take off', despite being an open source project."
And he managed to plug Jini and Java quite a bit as well, I see. I wish there was a transcript or audio file of the event.
#1 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 3:53 PM
Let's see if Sun's HotJava 3.0 takes off once it goes under the Sun Community source license... My bets on mozilla-heck we've already gotten companies to use mozilla in their products
#2 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 4:05 PM
Hehe, he even mentions Eiffel in there. I learned Eiffel maybe he should try it before calling it quirky. I would call it seriously screwed up in some places.
Personally, I would be very impress ed if anyone get the Mozilla source code smaller by 5 - 10%. For what it does its does and complies too its pretty good.
Could he have been taken out of context? Maybe he was just echoing what JWZ said about it being too complex.
Let's a face it, most of the important code for Mozilla is written by outside people.
The rendering engine code is just too complex for your average hobbyist open-source programmer to hop into.
It's not like adding a small module to Apache, or a command line option to GNU TAR.
This isn't an indictment of Mozilla, it applies equally to HotJava, IE, Opera, etc. Browsers are non-trivial apps unlike most open-source hacks.
Err,I meant most important Mozilla code was written by Netscape employees.
#5 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 6:42 PM
This is what I was talking about earlier regarding the timeframe of Mozilla. By the time it's released in final form, businesses will be too heavily vested in MSIE and will need a better reason to switch other than it's not from M$.
#6 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 7:05 PM
I don't know if I agree with that. IE has had over a year of time to penetrate the market. However, what we find is that with the advent of the Mozilla rendering control, app developers are opting to add the Mozilla renderer *in addition* to IE's. Homesite and TopStyle are just two examples. However, Mozilla offers much more in the area of extensibility: for example, the renderer used in the browser is the same used in composer. This will offer the ability for WYSIWYG development apps to incorporate a standards-compliant rendering engine inside their application - the same one that will be used by significant population of Internet surfers.
I wouldn't count Mozilla out yet. IE has not made as significant a penetration of the browser market as expected, despite their OS/browser bundling. In fact, Netscape is gaining in the business-user area.
#7 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 7:28 PM
alot of people like Netcenter too. Don't forget IM and ICQ. That's all mozilla territory
#8 Re:Bill Joy of Sun Spreads Little Joy
Thursday May 6th, 1999 8:25 PM
There's also the fact that MSIE 5 is turning out to be a major disappointment... Not much new there, other than the version number and a few new bits of tinsel.
#9 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 12:20 AM
I just wanted to respond to the guy who says that something like a rendering engine is too complex for the "average hobbyist open-source programmer." I believe NOTHING is too complex for anyone. If you want proof, just look at the Mnemonic project; they seem to have their layout engine mostly coded already, too. I am not meaning to bash Mozilla; I just think it is possible for it to have more involvement, and for its sake, it **NEEDS** that involvement. So don't dismiss outside effort as futile by saying that it is too complex for the hobbyist to understand.
#10 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 12:55 AM
I have a question for you: What happens when Win2000 and Office2000 are released?
We both know that millions upon millions of people are vested in Windows... so what then? What reason would one have to use a different browser technology other than the one already integrated into their two most heavily used software products (Windows & Office)?
I would really like to hear the other side of the coin on this issue because I want more than anyhting for Mozilla to be a complete and total success.
#11 Re: Bill Joy of Sun Spreads Little Joy
by Adam Lock <firstname.lastname@example.org>
Friday May 7th, 1999 3:02 AM
For the past year Mozilla has been broken so badly with such a large codebase that it's little wonder that few volunteers have come on board. The code is frankly scary, for someone who doesn't know COM/XPCOM. I am one of the few external developers at present.
Things are picking up though and this is because Mozilla has moved from broken to semi-working. I'm even using it now.
This is always how things happen on an opensource project;- first a handful of developers spend an age developing something genuinely useful to a semi-working state; then the volunteers start to appear, turning from a trickle to a flood as the product moves from alpha to GA quality.
#12 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 3:30 AM
By the way, Adam...
Excellent job on the npmozctl!
A few people in my neck of the woods have been whining about the UI wrapper included with Mozilla builds so I've been working on a little 'Communicator' style win32 wrapper for your control... it's coming along nicely... none of which would have been possible without the work you have done.
OpenSource projects as someone above said are getting developed for many many years and then suddenly, like linux, they take off (if they take of). Mozilla will probably take off very soon DESPITE being open source.
Speaking of HotJava, I recently had the chance to check out the documentation for their latest version (3.0), and just now its features are comparable to Netscape 3.0. NN 3.0 came out in 1997?
Keep up the good job, Sun :)
#15 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 4:59 AM
The problem is that people said just the same thing with OS/browser integration, and consumers *still* download Communicator.
I really don't see Win2000 and Office2000 as much of a problem. Why? Well, IE is already integrated in Win98, the OS that will be used by the most people (Win2000 won't - it's getting bad reviews and it's not even released yet, it's even bigger than Win98, and Linux is at this point the only OS that is gaining market share).
In addition, with OS/browser integration, the OS and browser are both only as strong as their weakest link - in this case IE (I have IE5 on my machine, in case you think my opinions of IE5 are made up). IE is not only unsecure (and thus makes the OS unsecure), but unstable and not standards compliant (it crashes more than Communicator, and brings the whole OS down with it). Also, the install for IE was *a complete joke*. I had to run the install 5 times to get IE isntalled. What the hell is that?
IE is losing ground in businesses (there were reports recently of IE gaining ground in consumer apps, but they had to admit that it was losing ground in the business sphere), and Windows, because it's not gaining ground, is losing as well.
Also, its way to early to know what will happen to the web application market because of XUL. I think we're going to see some impressive changes occur in the web application community (especially since XUL is built completely around standards).
#16 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 11:47 AM
Well, you brought up a good point. There are open-source browsers like Mnemonic and Konqueror, but they are no better than Mozilla, and actually worse as far as rendering engines go.
If open-source (real open-source, not corporate backed OSS) ever gets a complete CSS compliant browser, it will be because of one or two super smart individuals, who probably used to work at Adobe or Spyglass or in the SGML industry, and decided to start a project and seed the code.
But, if you take a look at the rendering engines in Jazilla/Jozilla/Mnemonic/Konqueror they all suck (slower and incorrect rendering)
A CSS rendering engine has a *FAR* higher cyclomatic complexity than an average application, and the number of test cases to get full path and statement coverage is astronomical.
#17 Re:Bill Joy of Sun Spreads Little Joy
by Adam Lock <email@example.com>
Friday May 7th, 1999 11:56 AM
OLEVariant - Thanks for your support!
Check out the CBrowse app in the tests folder of the ActiveX project. This is an app written in C++ and MFC that runs both the IE and mozilla controls. Apart from the initial CoCreateInstance, both controls run the same code.
CBrowse has grown from a simple applet into a test bed/harness. And has a quite a lot of reusable code in it.
#18 Re:Bill Joy of Sun Spreads Little Joy
Friday May 7th, 1999 3:19 PM
To the ones who thinks that complicated software can't be liberated (or Open Sourced as some here like to call it). I only say, gcc (the GNU C/C++/Fortran/ObjC Compiler) and Linux kernel. Two of the most complicated softwares out there.
Chief Scientist Bill Joy of Sun Microsystems Inc. criticized Netscape Communications 'Mozilla' effort for it's inability to 'take off', despite being an open source project."
from a bugzilla search, whether or not irrelevant for a wide group of people, Chief Scientist Bill Joy did not help Mozilla take off himself -- however
appear in a query on sun.com, e.g. Sun Microsystems? Now if more than two or three folks compile under Solaris and on Sun hardware and take interest in doing any kind of support for a open source browser in general, products Sun sell will include additional value worth buying.
GCC is an inferior compiler to most commercial compilers and if it were not for Cygnus, it would suck worse. (BTW, most of the outside GCC contribution was merely bug fixes or ports, not adding a new global optimizer, or making the C++ optimizer better, etc)
I will point out however, that compiler theory is TAUGHT at every university's CS department, and at good CS schools, most students are required to implement a compiler.
Actually, writing a compiler (non-optimizing) is fairly trivial. The parser piece is even more trivial.
Additionally, there is ample research on the subject from well known sources such as The DRAGON Book.
What colleges teach course geared to MAKRUP based rendering sengines (I'm not talking about graphics classes that teach raycasting, polygon rasterization, etc)
The fact is, writing someone thing like a CSS rendering engine, or a PDF viewer is non-trivial to get right, AND, is purely based on industry experience.
There is no "intro to text rendering" courses, no courses on how to implement a Quark clone, etc.
The final mark against GCC is simply HOW LONG it took to get to this stage.
If Open-Source takes 10 times as long to implementing something I can get commercially much sooner, I'll pay for it because my time is worth it.
By the way, the Linux kernel is not "complex", that's one of the major features of Linux is its simplicity compared to Windows maze-of-code.
And of course, Operating Systems development is also taught in all major universities. Indeed, Unix or Unix like kernels are being written all the time in universities by students.
There is a vast difference between a highly modularized system like Linux, where someone can write TAR without infering with you writing your SED utility.
A rendering engine has much higher intradependency.
Just try running gecko through a software metrics program and compare it to GCC or Linux's kernel's metrics.
I can't believe you people are pissed and in denial about a true statement. The fact of the matter is, Mozilla hasn't had as much outside contribution as other OSS projects.
#21 Re:Bill Joy of Sun Spreads Little Joy
Saturday May 8th, 1999 3:33 AM
Well, mozilla is complex and there are not 10000000 of outside developers but who cares, if the current developers keep up and finally release a working browser we have won, and I can't see anything that would stop it, I don't think AOL would pull the developers or microsoft take over the universe before mozilla can get finished.
The real threats come after mozilla gets released, Then we can get lots of hostile code forks and splittering. But hey, its you're lisence not mine.
#22 Browsers Are Complex
by -=Yusuf=- <firstname.lastname@example.org>
Saturday May 8th, 1999 4:11 PM
Yes, browsers are complex. A browser has to do lots of things, and with the Aurora project, it'll do more. Yes, it's non-trivial. Yes, you're right to an extent, BrowsersAreComplex.
But you're wrong where you imply that they are too complex for the dedicated programmer to comprehend.
NOTHING, read-my-lips, NOTHING is too complex to be understood with a bit of effort. Think back to your school days, and learning your first language, whatever it was. It was hard. You didn't come into the world knowing what a pointer was and what to do with it. But you learn.
And that's a really inadequate example, because learning how to code Mozilla will take more time than learning about pointers!! But the analogy stands, a person can learn it it he puts his mind to it. There's no lack of example code, go and download it. And there's no lack of people to help you, email Adam Lock (I'm sure he won't mind a FEW enquiries) or people @netscape.com.
Give it a few weeks, or months as the case may be and eventually you'll understand enough (assuming you're at a high enough level in basic C++ to start going over the code in the first place, I'm still doing Pascal.. =( ).
And even if non-industry related people cannot code it (a hard thing to believe), there are plenty of industry coders who will contribute to the smallest, fastest, coolest Lizard on the black as soon as they see the beta, I assume.
And THAT, ladies & gents, is the Inspirational Thought For The Day(TM) =)
PS oh and your point about the Open-source project taking 10 times as long? I'd use the commercial one too, but AS SOON AS the opensource came out I'd switch, as I'd get a better program. So would you, I'd like to assume....
since some mentioned mnemonic, I went to their site and the following disturbed me: <http://www.mnemonic.brows…org/todo/detail.php3?Id=2>
"The current code creates a new coordinate system for each block level element. (non-inline element, to be exact) We are making it different than CSS2, and this might pose a problem for dynamic-HTML"
In what ways does GCC "suck"? I'm using Visual C++ 6 at work and it has no support for some things that EGCS (now gcc) does, such as the try-function-block, and the relaxation of virtual function return types. There may be more, but I'm not sure.