Er... you need to read up on exception handling in Java... GC is about scope/reachability... try-finally is about exception handling. Exiting a finally MAY cause things to go out of scope and be available for GC.. or may not. In GC languages destructors don't really exist (finalize is kinda/sorta like one) so I suppose that is pretty mutually exclusive.
C++ doesn't have a finally since resources used by a class should be freed by the class in the destructor. Assuming you call the destructor then things will get cleaned up.
I'm not saying there's anything wrong with the parent posters knowledge of C++... rather they are bagging on GC when their arguments make it appear that they don't really know much about GC languages.
Still, their bug had nothing to do with GC and everything to do with a programming error that would have happened in any lang. I suppose you could argue that if they had used C++ they'd have found it due to running a profiler in the first place. But I'm not sure that's a positive one...
Hum... the whole thing kinda makes you wonder why they didn't profile it a long time ago anyway... but we've all done something similar at some point.
Unfortunately parent and the parent-parent have clearly never actually written anything of significance with a GC language. Using and try-finally have little to do with GC btw.
You can program sloppily in any language. If you knew exactly when objects were destroyed in C++ you'd never need a profiler. Clearly this is not the case for any non-trivial application (is sure as hell isn't mind-numbingly easy). Are you seriously saying smart pointers are easier to use than built in GC?
The advantage of GC languages is that for most cases you simply don't have to worry about memory management. It's a whole class of problems you just don't deal with. Sure, there are corner cases that'll get you, but once your familiar with them you can usually avoid them. And those corner cases require a lot less effort to avoid then all the various ways you can screw things up in C++. GC tech is pretty mature and while it's not suited to everything you've got yourself a big hill to climb if you want to argue that manual memory management will result in less buggy programs than GC.
Or was the website strange enough that I was inherently unwilling to give them ANY of my personal info, let alone my CC number?
I like the band, and I am willing to wade through any number of website disasters that reflect a given bands 'taste'. But when it comes to giving my money to someone I'd like something nice, normal and boring.
But there's no reason to use java as a scripting lang that you would need to have any awt/swing services. The dom and browser already do that for java script. The browsers could just treat java the same...
Heck...Even better would be something like groovy....
Parts still are. But they plan on fixing this as soon as they can.
Still point about using java makes a lot of sense regardless. You don't NEED awt/swing support to replace javascript. Just a context that the code in the page can run in that includes resources like the DOM, connection to source server, cookies repose, data cache, etc. Throw in the built in sandboxing and it's a big step forward.
Non trivial project, but then so is redoing javascript.
Work on a mac for 4 years like me and then try and use windows or linux. It's painful how haphazard and unrefined those UIs are, but I wouldn't ding them just for doing things the way they normally do them. (I would ding them for having endlessly nested dialogs for example).
Each of your issues is really more of an 'i don't like it' as opposed to 'the ui is horribly broken':
1. Only use the alt-tab (really command-tab)... There is only ONE instance of each app, then each app can have multiple 'document' windows. Tab needs to handle both.
2. Click one and hit return, edit name. Yeah, it's goofy, but the mac has done it that way forever...
3. OS X assumes you have multiple apps with multiple windows all going at the same time so you don't run things in full screen mode (cause then you'd have to alt-tab). Windows assumes you close an app to open another. Therefore full screen make more sense. Linux (KDE/Gnome) can run multiple apps, but has been built like windows. On large monitors those green + buttons do work nicely. Web browsers max out their height but don't set their width to 20+ inches.
4. You haven't tried looking very hard have you? Still, my os x term works about as well as any gnome or kde term i've used.. some how I mange to live with out tabs.
There are issues with the mac GUI to be sure... but the ones you have outlined are not really in that list.
AJ, I was agreeing with you and (for the sake of the discussion) willing to accept your proposition that:
a) Plame was covert according to the CIA b) Still isn't relevant since the definition of covert is defined by other than the CIA.
But as you say it's still kind of a grey area, Fitz was investigating the leak like he was supposed to. Libby lied his ass off to the grand jury and...
Lying to a Grand Jury is a Crime.
So it doesn't matter about the underlying case (ask martha stewart). Libby's punishment was tailored to that crime by following the sentencing guidelines. You know, rule of law and all that... The punishment had nothing to do with the underlying investigation. If we start getting into drivers of the case then I think it just makes the whole mess even more sordid. I mean come on, we all know that the WH was simply trying to smear Wilson... it was such SOP that they didn't even get worked up about the whole covert/not-covert thing.
WRT the partial pardon: Bush's so called position is so unsupportable that it's hard to know where to begin... Bush refused to commute death for people with much more of a case for leniency (let alone straight out retrials)... but he steps in for a close underling... pretty lame. And being bush he does the worst possible thing, admit the guy is guilty and let's him off anyway. Oh well, I suppose that is at least being honest about the double standard.
All of this is useful (thanks!). And totally beside the point.
Libby was convicted, argued that he had a good case for an appeal and got turned down flat ( by rather conservative judge ). So he was headed to jail. Period.
Bush didn't say "Oh he's not guilty so I'm gonna let him go". Bush effectively said, "Yeah, he's guilty and Dick doesn't give a rats ass". He tried to split some hairs to not piss off the law-and-order republicans but just ended up doing something stupid (as usual).
But then I realized we were looking at every action around him as a source of danger... Expecting anything to happen. And for that short few minutes we knew what it was like to be Tony. Could be the FBI waiting to grab you, could be some hired killers edging their way toward your table and entire family, the whole damn place could explode, etc...
No nice tidy bow wrapped ending for us or Tony, just another rev of the same wheel.
If you can live with a centralized repository then the only major thing lacking in SVN is merge tracking. Super-parent poster is cranked up on something strong. SVN's tagging and branching model is a GOOD thing. Makes it al simple to understand and therefore something people will actually use. Dev branches are a perfectly good way to avoid contention over mainline changes.
Also it's stupid to compare a centralized SCM and a distributed one and then complain that the centralized one doesn't solve the same problems the distributed one does (sorry linus).
For 90+% of the projects out there SVN is perfectly fine. Another 5% could use perforce... some vanishingly small percent really need something fully distributed like GIT. IF you need it then it's worth the extra complexity. If not then you're just wasting time.
> These days, I don't care whether they're "fun" or not as long as they aren't a jerk. If they can do the job demonstrably, they're hired, period.
"Fun" has nothing to do with it. Being infantile and/or an ass is disqualifying no matter how skilled. We interviewed a guy with straight A's, high sat's, really very smart (even dressed nice). But he spent half the interview bagging on his co-workers and other companies he interviewed with. Everyone gave him a 'no hire' before we even discussed him. Poor guy had worked at a big company for his first 6 years of employment and hadn't learned some basic interview no-nos...
1) Even super geeks need to have some people skills to get by in most places. Simply putting on a clean pair of jeans , nice shoes and a shirt with buttons on it say's you at least give a slight crap about getting along with other people. I'll hire someone in a t-shirt for a startup. Never, for a biz like medical care.
2) There's BS and then there's BS. A little polishing is good, too much is as auMater says... a bad sign. You gotta know which is which. You can check the degree by just asking the registrar if the person graduated as they said. No transcript required.
4) Super important in a small biz. I'll take a somewhat less skilled person that's good to have a beer with over super geek who's as ass any day. Kind of the same thing as 3. Good personality, tech skills and the ability to share and play nice with others is ideal.
So I've been working on web/feed crawlers for the past year. We thought that we'd just use the nofollow flag to prune the crawl space down. It's a standard right?
BZZT!
Turns out MAYBE 5% of all the sites we crawl have ANY use of the nofollow flag (let alone correct use of it). Blogger doesn't even use it. So the nsfw idea is nice. But doomed.
The whole semantic web idea is doomed w/o tools that force people to add the semantics to the content. And people (as a whole) just don't have any reason to make the effort. Human nature trumps geek desire at every turn.
Excellent idea... for the first 10 min. Then no one can figure out what the hell is going on without finding the particular docs for your particular extension to String. Especially the poor sap that has to figure out your code 2 years from now.
Yes, Ruby... I'm looking at you.
Ahghg!
Please don't flame me!. I love Ruby, I really do! It's the bestest ever!
Yes they do, until their applications stop running. People aren't going to blame MSFT for their Google apps not working. They're going to blame Google. "It's Google's responsibility to make sure our stuff works on the MSFT platform. Not the other way around."
People want their computers to run fast and easy. Aside from that, there are very few people that care how that is accomplished. So, if MSFT ensures that their computers are doing just that, they will have happier customers.
Er... "people" (real people not geeks) will think that the "internet" or the "web" is broken. They won't blame anyone except themselves or the company that makes the computer. So it'll be Dell, Gateway, HP and Apple that get calls about 'the email' not working. Blurring the line between desktop apps/web apps isn't going to change that. So don't expect consumer behavior to be the deciding factor in who might come out on top...
> What "dynamic typing" lets you do is forget about data types.
No. It let's you not worry about them when you just want to get something done. You'll worry about them when you use a lib and pass in a type it's not expecting at runtime. Again, automated testing is important.
> Source code reduction is a very good thing.
But it's not the ONLY thing. Readability/maintainability are important too.
I'd rather work with a slightly more verbose codebase where I can easily track down exactly what types things are so I can figure out what is going on at compile time and not just at runtime.
Dynamic typing can cause many of the problems that you refer to IF you don't have automated testing and if you don't write the tests as you develop the classes. If you do test driven development (or just have tests!) you can avoid a lot of these issues. The thought is these tests are roughly equivalent to the type checking that compiler or IDE would do. Plus they are more domain specific that just strict type checking.
Dynamic typing allows you to get right to solving the problem at hand and not get hung up on types and checked exceptions (different subject, but similar ideas). But it's not clear (to me) that it really saves you anything in the longer term. So personally, I'm still in the 'let's use a good ide and strict type checking' camp. But even so it's clearly overkill for a lot of tasks and something lighter is nice.
Nicely reasoned and practical post. Very refreshing.
I think a basic issue is that both camps don't know what they don't know about the other camp.
If you buy into the MS world you have plenty of tools for building out a lot of functionality. Most linux geeks don't get this. 'Normal' people and biz types don't give a squat about optimizing a driver or compiling a kernel. THe tech is a means to an end and not an important part of day to day operations.
But people in the MS world miss out on the creativity that comes from the open source jungle. They look out of the tour-bus and see a big hairy mess. But outside those windows are systems on a cd that will run a file server for years w/o being touched, elegant development frameworks and direct contacts between developers and their end users that benefit everyone, and more. But you have to get out of the bus and spend some time with the natives to really get it.
Vista/OSX -
Based on the VPs responses to the questions I have to wonder just how unwieldy the development of windows (the platform) has become. It sure sounds like it's pretty much completely out of control with VPs defining new processes and day to day developers ignoring them. And we all know vista has be ejecting features left and right. I'm sure they'll get it out the door at some point. And it'll be flashy. But at some stage enough is enough and they need to start over. Until they do the basic and fundamental issues in the windows platform will remain and no amount of layering new abstractions on top will fix them.
OSX has some structural advantages that are going to let it keep moving faster than Windows: - Modular codebase. Keeping Darwin separate from OSX is a very smart move. - Much smaller development organization. Fewer really good devs is better than a lot of devs where the skill levels are diluted. - Basic user application installation methods is far, far better than the windows model. - Very little backwards compatibility issues. - Leadership. I don't know Steve Jobs... but Blamer is no Steve Jobs. (And Gates isn't really Gates anymore is he?)
Eclipse 3.1 has improving GUI tools. But they are still pretty rough. Netbeans gui tools are a bit better. Pretty much the best that I have used is JBuilder. Not open source though. The point is that GUI's in java are NOT the same as guis in VB or winforms or whatever. Swing hasn't been a big priority for sun and it shows in most of the tools for building apps. I've built swing guis for a living and I would dearly love something that worked even half as good as Delphi did 10 years ago. anyways.. this leads into an argument about weather swing/java (even swt) is a good thing to be building desktop apps in anyway.
I can't address your particular xml issues.. but it does/can work. I use eclipse all day and have zero problems with xml docs. In the past i've used JBuilder exclusively (since 1.0) and a little netbeans too.
One of the painful aspects of most open source 'products' (esply dev tools!) is you have to be proactive in getting things working. XMLBuddy and the webapps tooling for eclipse are pretty essential plugins.
My advice would be to forget the classes or even feeling that you have to go pure open source. Get a copy of jbuilder foundation and learn java first. Then worry about the open source based ides. I keep jbuilder foundation around for the gui designer and use eclipse for the rest of the coding.
If you need them, wear them. Your eye muscles will fatigue if you don't. If you're vanity is against the the glasses then go out and drop 400 clams on a pair of 'hip' ones from a specialty store. If nothing else they can help with your geek-cool persona. (There is such a thing right?) Since you are younger get your prescription checked regularly since your vision is going to change.
Ideal lighting is typical living room. Diffuse, somewhat lower light. Basically set up some floor lamps without any being in your direct line of sight. Adjust your brightness as appropriate and avoid having it too bright. I've noticed that most non-mac monitors are WAY to contrasty and people tend to have the brightness turned all the way up. Typical computer lab lighting is horrible and not to be emulated.
As you get older your corneas will get harder and your distance vision will actually start to improve (at the cost of your near vision of course). It sounds like the poster is too young for this... but others sound like they might be experiencing this.
But what do I know... My first machine was an Atari 800. Which seems to mean that I am way too old.
Slashdot Mirror
Er... you need to read up on exception handling in Java... GC is about scope/reachability... try-finally is about exception handling. Exiting a finally MAY cause things to go out of scope and be available for GC.. or may not. In GC languages destructors don't really exist (finalize is kinda/sorta like one) so I suppose that is pretty mutually exclusive.
C++ doesn't have a finally since resources used by a class should be freed by the class in the destructor. Assuming you call the destructor then things will get cleaned up.
I'm not saying there's anything wrong with the parent posters knowledge of C++... rather they are bagging on GC when their arguments make it appear that they don't really know much about GC languages.
TFA is a FAdvert...
Still, their bug had nothing to do with GC and everything to do with a programming error that would have happened in any lang. I suppose you could argue that if they had used C++ they'd have found it due to running a profiler in the first place. But I'm not sure that's a positive one...
Hum... the whole thing kinda makes you wonder why they didn't profile it a long time ago anyway... but we've all done something similar at some point.
Unfortunately parent and the parent-parent have clearly never actually written anything of significance with a GC language. Using and try-finally have little to do with GC btw.
You can program sloppily in any language. If you knew exactly when objects were destroyed in C++ you'd never need a profiler. Clearly this is not the case for any non-trivial application (is sure as hell isn't mind-numbingly easy). Are you seriously saying smart pointers are easier to use than built in GC?
The advantage of GC languages is that for most cases you simply don't have to worry about memory management. It's a whole class of problems you just don't deal with. Sure, there are corner cases that'll get you, but once your familiar with them you can usually avoid them. And those corner cases require a lot less effort to avoid then all the various ways you can screw things up in C++. GC tech is pretty mature and while it's not suited to everything you've got yourself a big hill to climb if you want to argue that manual memory management will result in less buggy programs than GC.
Now I'm going to go read TFA...
The ever classic article from Courtney Love goes over the whole scheme:
http://archive.salon.com/tech/feature/2000/06/14/love/
Or was the website strange enough that I was inherently unwilling to give them ANY of my personal info, let alone my CC number?
I like the band, and I am willing to wade through any number of website disasters that reflect a given bands 'taste'. But when it comes to giving my money to someone I'd like something nice, normal and boring.
But that's just me.
But there's no reason to use java as a scripting lang that you would need to have any awt/swing services. The dom and browser already do that for java script. The browsers could just treat java the same...
Heck...Even better would be something like groovy....
Parts still are. But they plan on fixing this as soon as they can.
Still point about using java makes a lot of sense regardless. You don't NEED awt/swing support to replace javascript. Just a context that the code in the page can run in that includes resources like the DOM, connection to source server, cookies repose, data cache, etc. Throw in the built in sandboxing and it's a big step forward.
Non trivial project, but then so is redoing javascript.
One of the highest ignorance to knowledge ratios in Slashdot's history...
90%+ of the c/c++ guys have no clue.
99%+ of those bagging on GC have less of a clue.
They might as well be talking about girls...
Geez.
You're having a context switching issues...
Work on a mac for 4 years like me and then try and use windows or linux. It's painful how haphazard and unrefined those UIs are, but I wouldn't ding them just for doing things the way they normally do them. (I would ding them for having endlessly nested dialogs for example).
Each of your issues is really more of an 'i don't like it' as opposed to 'the ui is horribly broken':
1. Only use the alt-tab (really command-tab)... There is only ONE instance of each app, then each app can have multiple 'document' windows. Tab needs to handle both.
2. Click one and hit return, edit name. Yeah, it's goofy, but the mac has done it that way forever...
3. OS X assumes you have multiple apps with multiple windows all going at the same time so you don't run things in full screen mode (cause then you'd have to alt-tab). Windows assumes you close an app to open another. Therefore full screen make more sense. Linux (KDE/Gnome) can run multiple apps, but has been built like windows. On large monitors those green + buttons do work nicely. Web browsers max out their height but don't set their width to 20+ inches.
4. You haven't tried looking very hard have you? Still, my os x term works about as well as any gnome or kde term i've used.. some how I mange to live with out tabs.
There are issues with the mac GUI to be sure... but the ones you have outlined are not really in that list.
(why can't I help myself?)
c le/2007/07/02/AR2007070201611.html?hpid=opinionsbo x1
3 02407
7 _07_07.shtml#1183437010
c le/2007/07/02/AR2007070202060.html?hpid=topnews
The freaking WP editors think it was a boneheaded move:
http://www.washingtonpost.com/wp-dyn/content/arti
25 year diabetic vet doesn't get a lighter sentence for pretty much the same thing as libby:
http://www.abcnews.go.com/Politics/wireStory?id=3
The entire investigation is by republicans for republicans (yet it's a liberal witch hunt party!)
http://volokh.com/archives/archive_2007_07_01-200
Dick's fingers are all over this one (eww!)... Justice Department Guidelines? Go F___ yourself!
http://www.washingtonpost.com/wp-dyn/content/arti
AJ, I was agreeing with you and (for the sake of the discussion) willing to accept your proposition that:
a) Plame was covert according to the CIA
b) Still isn't relevant since the definition of covert is defined by other than the CIA.
But as you say it's still kind of a grey area, Fitz was investigating the leak like he was supposed to. Libby lied his ass off to the grand jury and...
Lying to a Grand Jury is a Crime.
So it doesn't matter about the underlying case (ask martha stewart). Libby's punishment was tailored to that crime by following the sentencing guidelines. You know, rule of law and all that... The punishment had nothing to do with the underlying investigation. If we start getting into drivers of the case then I think it just makes the whole mess even more sordid. I mean come on, we all know that the WH was simply trying to smear Wilson... it was such SOP that they didn't even get worked up about the whole covert/not-covert thing.
WRT the partial pardon: Bush's so called position is so unsupportable that it's hard to know where to begin... Bush refused to commute death for people with much more of a case for leniency (let alone straight out retrials)... but he steps in for a close underling... pretty lame. And being bush he does the worst possible thing, admit the guy is guilty and let's him off anyway. Oh well, I suppose that is at least being honest about the double standard.
But that's just my biased opinion.
All of this is useful (thanks!). And totally beside the point.
Libby was convicted, argued that he had a good case for an appeal and got turned down flat ( by rather conservative judge ). So he was headed to jail. Period.
Bush didn't say "Oh he's not guilty so I'm gonna let him go". Bush effectively said, "Yeah, he's guilty and Dick doesn't give a rats ass". He tried to split some hairs to not piss off the law-and-order republicans but just ended up doing something stupid (as usual).
sigh
At first I was like: WTF?
But then I realized we were looking at every action around him as a source of danger... Expecting anything to happen. And for that short few minutes we knew what it was like to be Tony. Could be the FBI waiting to grab you, could be some hired killers edging their way toward your table and entire family, the whole damn place could explode, etc...
No nice tidy bow wrapped ending for us or Tony, just another rev of the same wheel.
Thank you!
If you can live with a centralized repository then the only major thing lacking in SVN is merge tracking. Super-parent poster is cranked up on something strong. SVN's tagging and branching model is a GOOD thing. Makes it al simple to understand and therefore something people will actually use. Dev branches are a perfectly good way to avoid contention over mainline changes.
Also it's stupid to compare a centralized SCM and a distributed one and then complain that the centralized one doesn't solve the same problems the distributed one does (sorry linus).
For 90+% of the projects out there SVN is perfectly fine. Another 5% could use perforce... some vanishingly small percent really need something fully distributed like GIT. IF you need it then it's worth the extra complexity. If not then you're just wasting time.
Exactly my position:
> These days, I don't care whether they're "fun" or not as long as they aren't a jerk. If they can do the job demonstrably, they're hired, period.
"Fun" has nothing to do with it. Being infantile and/or an ass is disqualifying no matter how skilled. We interviewed a guy with straight A's, high sat's, really very smart (even dressed nice). But he spent half the interview bagging on his co-workers and other companies he interviewed with. Everyone gave him a 'no hire' before we even discussed him. Poor guy had worked at a big company for his first 6 years of employment and hadn't learned some basic interview no-nos...
1) Even super geeks need to have some people skills to get by in most places. Simply putting on a clean pair of jeans , nice shoes and a shirt with buttons on it say's you at least give a slight crap about getting along with other people. I'll hire someone in a t-shirt for a startup. Never, for a biz like medical care.
2) There's BS and then there's BS. A little polishing is good, too much is as auMater says... a bad sign. You gotta know which is which. You can check the degree by just asking the registrar if the person graduated as they said. No transcript required.
4) Super important in a small biz. I'll take a somewhat less skilled person that's good to have a beer with over super geek who's as ass any day. Kind of the same thing as 3. Good personality, tech skills and the ability to share and play nice with others is ideal.
So I've been working on web/feed crawlers for the past year. We thought that we'd just use the nofollow flag to prune the crawl space down. It's a standard right?
BZZT! Turns out MAYBE 5% of all the sites we crawl have ANY use of the nofollow flag (let alone correct use of it). Blogger doesn't even use it. So the nsfw idea is nice. But doomed.
The whole semantic web idea is doomed w/o tools that force people to add the semantics to the content. And people (as a whole) just don't have any reason to make the effort. Human nature trumps geek desire at every turn.
Excellent idea... for the first 10 min. Then no one can figure out what the hell is going on without finding the particular docs for your particular extension to String. Especially the poor sap that has to figure out your code 2 years from now.
Yes, Ruby... I'm looking at you.
Ahghg!
Please don't flame me!. I love Ruby, I really do! It's the bestest ever!
Really.
Ahem...
Er... "people" (real people not geeks) will think that the "internet" or the "web" is broken. They won't blame anyone except themselves or the company that makes the computer. So it'll be Dell, Gateway, HP and Apple that get calls about 'the email' not working. Blurring the line between desktop apps/web apps isn't going to change that. So don't expect consumer behavior to be the deciding factor in who might come out on top...
Let's see... be cold all the time and eat just enough to keep your body going.
Sounds like a ton of fun.
I think I'll pass...
> What "dynamic typing" lets you do is forget about data types.
No. It let's you not worry about them when you just want to get something done. You'll worry about them when you use a lib and pass in a type it's not expecting at runtime. Again, automated testing is important.
> Source code reduction is a very good thing.
But it's not the ONLY thing. Readability/maintainability are important too.
I'd rather work with a slightly more verbose codebase where I can easily track down exactly what types things are so I can figure out what is going on at compile time and not just at runtime.
Dynamic typing can cause many of the problems that you refer to IF you don't have automated testing and if you don't write the tests as you develop the classes. If you do test driven development (or just have tests!) you can avoid a lot of these issues. The thought is these tests are roughly equivalent to the type checking that compiler or IDE would do. Plus they are more domain specific that just strict type checking.
Dynamic typing allows you to get right to solving the problem at hand and not get hung up on types and checked exceptions (different subject, but similar ideas). But it's not clear (to me) that it really saves you anything in the longer term. So personally, I'm still in the 'let's use a good ide and strict type checking' camp. But even so it's clearly overkill for a lot of tasks and something lighter is nice.
Nicely reasoned and practical post. Very refreshing.
I think a basic issue is that both camps don't know what they don't know about the other camp.
If you buy into the MS world you have plenty of tools for building out a lot of functionality. Most linux geeks don't get this. 'Normal' people and biz types don't give a squat about optimizing a driver or compiling a kernel. THe tech is a means to an end and not an important part of day to day operations.
But people in the MS world miss out on the creativity that comes from the open source jungle. They look out of the tour-bus and see a big hairy mess. But outside those windows are systems on a cd that will run a file server for years w/o being touched, elegant development frameworks and direct contacts between developers and their end users that benefit everyone, and more. But you have to get out of the bus and spend some time with the natives to really get it.
Vista/OSX -
Based on the VPs responses to the questions I have to wonder just how unwieldy the development of windows (the platform) has become. It sure sounds like it's pretty much completely out of control with VPs defining new processes and day to day developers ignoring them. And we all know vista has be ejecting features left and right. I'm sure they'll get it out the door at some point. And it'll be flashy. But at some stage enough is enough and they need to start over. Until they do the basic and fundamental issues in the windows platform will remain and no amount of layering new abstractions on top will fix them.
OSX has some structural advantages that are going to let it keep moving faster than Windows:
- Modular codebase. Keeping Darwin separate from OSX is a very smart move.
- Much smaller development organization. Fewer really good devs is better than a lot of devs where the skill levels are diluted.
- Basic user application installation methods is far, far better than the windows model.
- Very little backwards compatibility issues.
- Leadership. I don't know Steve Jobs... but Blamer is no Steve Jobs. (And Gates isn't really Gates anymore is he?)
Eclipse 3.1 has improving GUI tools. But they are still pretty rough. Netbeans gui tools are a bit better. Pretty much the best that I have used is JBuilder. Not open source though. The point is that GUI's in java are NOT the same as guis in VB or winforms or whatever. Swing hasn't been a big priority for sun and it shows in most of the tools for building apps. I've built swing guis for a living and I would dearly love something that worked even half as good as Delphi did 10 years ago. anyways.. this leads into an argument about weather swing/java (even swt) is a good thing to be building desktop apps in anyway.
I can't address your particular xml issues.. but it does/can work. I use eclipse all day and have zero problems with xml docs. In the past i've used JBuilder exclusively (since 1.0) and a little netbeans too.
One of the painful aspects of most open source 'products' (esply dev tools!) is you have to be proactive in getting things working. XMLBuddy and the webapps tooling for eclipse are pretty essential plugins.
My advice would be to forget the classes or even feeling that you have to go pure open source. Get a copy of jbuilder foundation and learn java first. Then worry about the open source based ides. I keep jbuilder foundation around for the gui designer and use eclipse for the rest of the coding.
If you need them, wear them. Your eye muscles will fatigue if you don't. If you're vanity is against the the glasses then go out and drop 400 clams on a pair of 'hip' ones from a specialty store. If nothing else they can help with your geek-cool persona. (There is such a thing right?) Since you are younger get your prescription checked regularly since your vision is going to change.
Ideal lighting is typical living room. Diffuse, somewhat lower light. Basically set up some floor lamps without any being in your direct line of sight. Adjust your brightness as appropriate and avoid having it too bright. I've noticed that most non-mac monitors are WAY to contrasty and people tend to have the brightness turned all the way up. Typical computer lab lighting is horrible and not to be emulated.
As you get older your corneas will get harder and your distance vision will actually start to improve (at the cost of your near vision of course). It sounds like the poster is too young for this... but others sound like they might be experiencing this.
But what do I know... My first machine was an Atari 800. Which seems to mean that I am way too old.