Good day,
Javalobby’s Rick came up with this idea today: http://www.javalobby.org/nl/archive/jlnews_20031216o.html
What do you think? If I remember correctly from reading this forum there are a few Java bugs which tend to disturb game orientated developers…
The one drawback that I see is that people will tend to fix the bugs in the context of how it effects them or the platform that they use. Java is complex… How many of us have Solaris to test on? How many of us will run our apps in Hebrew to make sure the text rendering is OK? Do we have the same security concerns & awareness that someone else might or will we open a security hole when we fix something?
Sun engineers know the Java code base and how things fit together much better than outsiders. As long as the community submitted fixes are properly analyzed by the people that know best then yes I can see this work. I don’t want Java to become as unstable and hacked up as some other open Source wonders cough Linux **cough **
As an example of how you need the inside knowledge to do things right, I challenge anyone to find an ICQ client not made by the ICQ folks that works the same as the ‘official’ ICQ client.
Every one that I have tried (and I have tried a few) fails in rather significant ways one way or another. Eg. messages over 450 chars, showing proper online status, file transfers, simply sending and receiving messages reliably, downloading contact lists from the server, simply crashing frequently, etc. Every 3rd party ICQ client fails in at least one of these ways.
Nobody would want chunks of Java to become broken or unreliable in the same way, simply because people fixing bugs didn’t have a good grasp of the whole picture.
The difference, of course, being that the 3rd party ICQ developers don’t have access to the original source code.
Yes, but they can watch the data being sent when the real ICQ client is being used… so they should know when they are sending something different…
It is different, but the point is that the people that work on the project for a living are usually in a MUCH better position to fix things without breaking it.
Reverse engineering a protocol is much harder than seeing the actual code for it.
The source code for java is available, and if someone spots and fixes a bug, that’s a Very Good Thing.
Sun can still run the patch through their tests and internal quality controls if they want to, but I really don’t see a problem with this.
[quote]As long as the community submitted fixes are properly analyzed by the people that know best then yes I can see this work.
[/quote]
I think this could be one way, yes.
[quote]I don’t want Java to become as unstable and hacked up as some other open Source wonders cough Linux **cough **
[/quote]
If I remember correctly the total-up-time of Linux is much better than the one of Windows NT (NT4, 2000, XP). I’m not sure it’s fair to say Linux was unstable. To me the opposite has been true, but then I’ve never worked with Macos-X, unfortunately.
You’re a Macos-X user so isn’t your nice Macos-X basing on such an “Opensource wonder”, too? 
To your ICQ example: Personally I’ve made bad experiences with the original ICQ client (together with some collagues) and switched years ago to an Opensource ICQ clone (Miranda, small and smart).
However I think this is another topic.
Back to the topic: One of Rick’s main points have been that several long-term Java bugs are there since many years and you can say they won’t be fixed by SUN. If it’s one bug which hinders you from using Java for your purpose there is some need to fix it.
Are the business Java bugs being discussed in a good way in the Javalobby forum? If so, we could maybe concentrate on the game related Java bugs which need to be fixed.
[quote]If I remember correctly the total-up-time of Linux is much better than the one of Windows NT (NT4, 2000, XP).
[/quote]
Perhaps in a server capacity, though the Windows servers at work have only had hardware related failures. As a desktop system MOST Linux software remains very fragile in my experience.
[quote]I’m not sure it’s fair to say Linux was unstable. To me the opposite has been true, but then I’ve never worked with Macos-X, unfortunately.
[/quote]
I’m just basing it on my own experience trying several linux versions on a few machines over the last 3 years. Linux is starting to look better and people are actually making UIs for things finally… but fundamentally it is VERY unpolished.
[quote]You’re a Macos-X user so isn’t your nice Macos-X basing on such an “Opensource wonder”, too?
[/quote]
Yes, I guess so, it is BSD unix built on the Mach kernel developed at CMU (the same kernel the was the basis of the Windows NT/2k/XP kernel)… the apps and UI though are more stable and MUCH better designed. Thankfully the display is not based on X windows and the UI is not trying to be a Windows clone that doesn’t work as well.
[quote]To your ICQ example: Personally I’ve made bad experiences with the original ICQ client
[/quote]
Yes IT DOES suck - the problem is that all the others, while trying to improve some things break what does work on the original[quote] …switched years ago to an Opensource ICQ clone (Miranda, small and smart).
However I think this is another topic.
[/quote]
Tried Miranda recently- it doesn’t work properly. But you are right this is all another topic that doesn’t even belong in these boards anyway…
[quote]Back to the topic: One of Rick’s main points have been that several long-term Java bugs are there since many years and you can say they won’t be fixed by SUN.
[/quote]
We have to ask ourselves… why? I do not believe that Sun is being malicious and is purposely not fixing the bugs. Rather I think what appears to be a simple bug is not so simple because of factors that may not be obvious to the community. Eg. the fix will break something somewhere that we aren’t aware of. In some cases it is best to simply live with the bug and let it be known and find alternative solutions. This many mean deprecating a troublesome API, but that isn’t always possible.
[quote]…we could maybe concentrate on the game related Java bugs which need to be fixed.
[/quote]
By all means I think it is right to suggest fixes to Sun… obviously more so when it effects our efforts directly. It can’t hurt to try… as long as the fixes are properly evaluated.
The problem is just having anyone fix bugs, without the benefit of seeing the entire body of Java technology, creates instability. A bug fix here may cause 3 more to appear there, and may not make them known in that person’s app. As well, for any bug to be badged as “fixed” means it has to go through a whole bunch of testing, and these test environments are not available to everyone (for the purpose of maintaining a common code platform without people messing with the testing enviroments).
-ChrisM
There is no way on earth I would submit a bug fix to a Sun product (or an MS product for that matter) even if I had the skill and know-how. The company I work for charges clients $155 an hour for programmer time(I don’t make near that of course). I know what my time is worth, and it sure as hades isn’t free.
Now on the other hand an opensource project is another matter entirely because you know what your getting into. Java is very frustrating to use because of all the little inconsistancies in implementations, but it is still the only write-once-using-obscure-hacks-to-accomade-different-environments, and run-anywhere (almost) thing going.
[quote]There is no way on earth I would submit a bug fix to a Sun product (or an MS product for that matter) even if I had the skill and know-how. The company I work for charges clients $155 an hour for programmer time(I don’t make near that of course). I know what my time is worth, and it sure as hades isn’t free.
Now on the other hand an opensource project is another matter entirely because you know what your getting into. Java is very frustrating to use because of all the little inconsistancies in implementations, but it is still the only write-once-using-obscure-hacks-to-accomade-different-environments, and run-anywhere (almost) thing going.
[/quote]
I find that argument a bit twisted. With Open Source you know what you are getting into only in the sense that your hard work will also benefit others and you won’t be paid. Given that you are essentially giving away your code in either case, that you will be benefiting developers in either case, etc. I find your argument very unconvincing. My time is worth money too, that doesn’t stop me from helping out here or anywhere else.
So if you knew how to fix a bug in the JRE and you had the fix you would hold it for ransom rather than see the bug fixed? To what end?
P.S. There are none to very few obscure hacks needed to write perfectly decent, complex, cross-platform Java programs. I’ve come from a background of writing for Windows and trust me there are a LOT more inconsistencies and obscure hacks required to get stuff to work on that single platform when programming in C/C++.
A lot of naive people (and personally I would not be surprised to find Ross among them) don’t realize what is involved in fixing bugs in a product as complex and mature as the V
A lot of naive people (and personally I would not be surprised to find Ross among them) don’t realize what is involved in fixing bugs in a product as complex and mature as the VM.
The majority of time is not spent in code development but in testing to be absolutely assured that a fix does not cause a regression or new bug. In addition to functionality tests we also do nightly performance tests for the same reasons.
Having people just throw their idea of a fix over the wall turns out not to be all that useful and can in fact slow development down if a great deal of testing time gets taken up testing “fixes” that have to be rejected.
So its a multi-edged sword, really.
Then again if this diverted the energy Ross spends spouting off to something useful like code development it couldn’t be ALL bad…
meow
And even so, historically there has been nowhere near enough regression and unit testing…
Of the 15-20 bugs in Java I’ve had accepted (and most fixed by now), about 50% were things that would have been picked up by a basic unit test.
FYI decades of research into software development have concluded that over the lifetime of any long-lived product, testing takes more time and more money than everything else put together. If the product is really long-lived, everything else becomes so small as to become entirely irrelevant in terms of profit etc. A product that lasts 2 decades can quite easily have sunk 90% of cost into testing (and resulting fixes…)
Sun is doing pretty well in this regard, despite frustratingly obvious bugs that I’ve come across (and which point to some developer somewhere who ought to be moved into the marketing dept - or somewhere else where they aren’t allowed to code!) - getting everyone across an organization to write good tests to say 95% coverage is extremely difficult; policing it is almost impossible :(. I’m just pointing out that even knowing the problems (of regression testing - of which unit testing is a major part), and with lots of effort on solving them, lots still falls through the gaps. It’s really hard.
Popular (i.e. with lots of developers) open source software almost universally sucks quality-wise, usually because of a laissez-faire attitude to this issue, or just plain old rubbish developers who pretend to add decent unit tests but don’t really. There are exceptions, although the only ones I can think of are enterprise software, where the main developers are people who’ve come from a corporate environment and know all this - so they re-produce the good practice (and are draconian about enforcing it) on their OS projects.
IMHO, Cathedral vs Bazaar is an irrelevancy when it comes to quality - it’s a philosophical argument that skirts the real issues (ducks and runs for cover ).
[quote]Then again if this diverted the energy Ross spends spouting off to something useful like code development it couldn’t be ALL bad…
[/quote]
Well, fair enough Rick does a good job usually and it’s good for SUN that there are so enthusiastic Java users out there, joining together at one central place/lobby.
Rick’s idea with the, let’s say, “help in fixing bugs” isn’t bad basically. It depends on how it will work.
In contrast to some posters in this thread my team and me have made very good experiences with small and large Opensource products at company and at home. Generally speaking the qualitity of Opensource products isn’t a single bit worse than that of Closed-source software. To my experience the contrary is true. This also has to do with the fact that much more people around the world look at the source code and indeed find bugs.
Swpalmer wrote[quote]As long as the community submitted fixes are properly analyzed by the people that know best then yes I can see this work.
[/quote]
I fully back up this statement and I think it’s well “compatible” with Rick’s basic points.
Sorry, got a little carried away on the obscure hack bit, I was thinking about fullscreen in Mac OS when I wrote that.
[quote] I find that argument a bit twisted. With Open Source you know what you are getting into only in the sense that your hard work will also benefit others and you won’t be paid. Given that you are essentially giving away your code in either case, that you will be benefiting developers in either case, etc. I find your argument very unconvincing.
[/quote]
The implied idea that I want to ‘benefit developers’ at the expense of my own interest is where you misunderstand my position.
The thing is, if I contribute to an opensource project I know I won’t be making money. I also know no one else will be making money from my work either.
To me, opensource software is play, I will share that. Commercial software is work, that pays my bills. I won’t do that for free.
[quote]The thing is, if I contribute to an opensource project I know I won’t be making money. I also know no one else will be making money from my work either.
To me, opensource software is play, I will share that. Commercial software is work, that pays my bills. I won’t do that for free.
[/quote]
Ah, you are choosing a particular OS license that does not allow commercial use.
If someone asks a question here and they are working on a commercial game are you not going to answer because they will profit from it without paying you? Not that there is anything wrong with that… heck there are places such as the (Sun endorsed) HotDispatch.com where that is entirely the point.
I do see your middle ground though: Sell it, donate it, or take a typical OS stance and donate it to people that don’t make money (can’t afford it), sell it to those that do (presumably it helps them make money so they can afford to give you your cut) - I think that is your position. I’m just saying that I have benefited quite a bit from Java without paying Sun a single penny, and it seems perfectly reasonable to give them a little bit back for free, otherwise I’m a moocher.
[quote]IMHO, Cathedral vs Bazaar is an irrelevancy when it comes to quality - it’s a philosophical argument that skirts the real issues (ducks and runs for cover ).
[/quote]
Hehe… I actually agree. There are a lot more crappy Os projects out there than good ones. Something that seems to be regard in the community as a positive thing is that it is so wonderful that you have the source and therefore can fix problems if you encounter them. The PROBLEM unfortunately is that this is true - when there are bugs YOU often have to fix them. There is often no party that has an obligation to fix them for you. With open source you get the source AND the bugs. If it’s broken hey no problem YOU can fix it. Thanks, just what I wanted, a nice bunch of code, thousands of lines of it, that I didn’t write but that I have to debug :). I’m presenting this rather extreme view simply to illustrate this point, in isn’t always as bad as I elude to here. Open Source isn’t all bad… but it is not nearly as great as the hype makes it out to be. Most of the time when I grab a library to do some function for me it is precisely because I want the functionality that it provides to be somebody else’s problem