The long answer is: Yes, it's a lot faster retreieving data. It's about as fast as looking up an object in an in-memory array, because that's what it does. However, they are comparing apples to oranges.
Prevayler is a very neat piece of software. It's very simple, and not hard to implement really. The problem is that the people behind Prevayler seems to think that it's the end-all-be-all of databases. It's not.
What these people fail to mention is that you lose somehting that a lot of people find very useful in SQL databases: select. Yes people, that's right. It doesn't provide any way of searching or joining tables. In fact, you don't even have tables. All that Prevayler really is, is a method to encapsulate all data modifications in an action object which is persisted to disk, so that they can be re-played whenever the system starts after a crash. All data accesess are performed just like any other memory read, in the same way as you access any other object. Of course it'll be faster than accessing an SQL database.
The Prevayler people are very good at twisting the truth to create some amazing benchmark results. It's a good technology, but the the attitude of the developers is slightly irritating.
Well, MS Office is the only swedish application I can think of that doesn't use Control-F for search. Say what you like, but the shortcuts are not consistent. It's all up to the developer.
And by the way, how many GNOME applications can you think of that doesn't use Control-C etc, for cupt&paste (unless you configure it otherwise).
Wow. I'm convinced now.
I never expected you to be. The intention was to convince other readers.
90% of applications follow standard shortcut procedure, but there are always the exceptions, which aren't the fault of Windows consistency.
Umm, Control-F makes bold text in the swedish version of MS Office, and starts a search in just about every other application. They can't even get it right within their own set of applications.
Completely false. Windows is considered a bastion of homogenized consistency (good or bad), especially compared to the hell that is the Linux desktop attempt.
If you install a stock RedHat 9 system and start using everything they ship, and, when you want to download new apps, make sure you download GNOME or GTK+ apps, you are looking at a pretty standardised user interface.
Before I installed RedHat 9 my GF was rebooting into windows all the time (I had this WindowMaker+assorted tools setup) becuase she just felt Windows was easier to use. After I reinstalled with RedHat 9 and decided to stop mess around I realised that she wasn't rebooting anymore. I didn't even ask her. She really felt it was nicer to use the GNOME desktop than Windows.
So, you can complain all you like about how crappy the Linux desktop is, but I have actual proof that at least one fairly computer ilitterate person prefers GNOME before Windows. (and no, she's no power user. she uses Mozilla, various text editors, Open Office, GIMP (for working with the digital photos) and I hardly ever have to help her).
Why is Windows so consistent in having Control-C etc... being cut&paste keys. They add a million "email-buttons" and other crap to the "windows keyboards" but they fail to add the thee most important buttons...
Cut, Copy and Paste buttons.
Sun has had them for 15 or so years. Fortunately I can cofigure my Sun keyboard I have connected to my Linux PC to use these buttons for copy&paste. I have yet to find a way to do it in Windows.
and if I found up that the "Internet Access" I signed up for had all of those caveats, I'd switch fast before I ran into other such nonsense.
Are you sure? If you didn't know about it, you'd most likely order it, and then you'd have to suffer a night (or a weekend) of not getting full Internet access until you could opt-out. Or, you knew about it and you'd have opted out at the same time you ordered the connection. In both cases, you'd still be using them. (and in this particular case I think you would, since they give you 10 MBit/s up and downstream:-) )
I honestly don't know where they provide this information. I had connected everything before they even installed the switches in the basement of out apartment building, and started using the stuff when I saw the lights on my switch activate.:-) As soon as I noticed they were blocking incoming traffic I opted out.
I still think that their way is the best. It will protect most network-illiterate users, and still open up everything for us who want it.
You can add one swedish provider that also blocks all incoming traffic until you opt-out (which involves signing a document relieving them from any responsibility if your machine is cracked).
Oh, and by the way: Even before I opted out of their firewall, I could play pretty much all online games (but not host). So I suppose very few people will even notice they have the firewall.
The Open Group says the following on the web page refernced to by the article:
Reference to the SCO web site shows that they own certain intellectual property and that they correctly attribute the trademark to The Open Group. SCO has never owned "UNIX". SCO is licensed to use the registered trademark UNIX "on and in connection" with their products that have been certified by The Open Group, as are all other licensees.Open Group
Wouldn't it be nice if The Open Group revoked SCO's license to use the UNIX trademark?
I guess that if a woman can get millions of dollars for spilling hot coffee on herself, someone else can get a refund for a CD that they couldn't play in their computer.
If she spilled the hot coffee, then obviously her computers cup holder is broken... No refund!
Yeah, it is a bit crap. I prefer the C# syntax (foreach... in...), but they were insistent on not adding new keywords in Java.
That didn't stop them from adding "assert" as a keyword in 1.4. Wonder what the rationale behind that was?
It's not exactly broken. In your example foo is not an object. When you assign to x and y you automatically create an object. It's just like you'd done:
Well, that was my point actually. A seemingly innocent assignment actually creates a new object, but only depending on what "foo" is actually defined as.
For example, what will the value of "bar" be after the following code?
x = 10; y = 10; boolean bar = (x == y);
In Java today, I'm 100% sure that bar will be true. After 1.5 (and in C#) I don't know this, it can be true or false, depending on wether x and y are Integer's or int's. Integer and int are not the same thing, and it should be evident from looking at the code. That's the orthogonality I was referring to.
Systems like xdoclet already do that, but it's a bit too ad hoc for me. I'm not a big fan of functionality hidden in comments. I should be able to strip all the comments from my code and have it compile to the same program and work the same way.
Well, look at it this way:/** is not really the comment character, it's the javadoc block start character. Adding special syntax do deal with metadata is really not nessecary, since xdoclet seems to get by without it. There are so many other things that could be fixed instead, for example making array objects and strings implement java.util.List.
It's my firm belief that these features (especially static imports) were added for no other reason than to shut up the worst C# whiners.
In my opinion, Java would be better off without many of those new features.
The enhanced for loop has a horrible syntax, which only saves a few keystrokes at the expense of being very hard to read.
The autoboxing breaks othogonality in object identification. Consider the following:
{ Integer foo = 10; ... Integer x = foo; Integer y = foo; if(x == y) { // objects are the same } }
{ int foo = 10; ... Integer x = foo; Integer y = foo; if(x == y) { // objects not the same? WTF?! } }
(how do I ebped code in a post and retain the leading spaces?)
C# programmers don't care about this very much because they already have this problem times 1000 because value objects and object references have the same syntax.
Type-safe enums are just a compile-time code generator that creates the same kind of type-safe enums Java programmers have been using for ages. However, being able to use them in swich/case statements is an avantage.
Static import is just plain silly. You save a few keystokes when calling static methods (something you do quite rarely anyway) at the expense of getting a lot more confusing code. The only situation where it might be nice to have is when you wring lots of mathematic expressions. I don't think this advantage outweighs the problems though.
Metadata: why not simply extend the javadoc syntax?
I have used both VS.NET and IntelliJ IDEA and in my experience, IDEA beats VS.NET on almost every point. In fact, VS.NET feels like a prison that prevents me from being efficient.
I'd be very happy to hear from someone who tried both, and still like VS.NET more.
I have been considering migrating to GnuCash for quite some time, but I still haven't succeeded in finding a way of converting my MoneyDance data to GnuCash. It aways fails with an "illegal date format" when reading the Quicken files exported from MoneyDance.
Slashdot Mirror
Nope, definately doesn't rhyme.
The long answer is: Yes, it's a lot faster retreieving data. It's about as fast as looking up an object in an in-memory array, because that's what it does. However, they are comparing apples to oranges.
Prevayler is a very neat piece of software. It's very simple, and not hard to implement really. The problem is that the people behind Prevayler seems to think that it's the end-all-be-all of databases. It's not.
What these people fail to mention is that you lose somehting that a lot of people find very useful in SQL databases: select. Yes people, that's right. It doesn't provide any way of searching or joining tables. In fact, you don't even have tables. All that Prevayler really is, is a method to encapsulate all data modifications in an action object which is persisted to disk, so that they can be re-played whenever the system starts after a crash. All data accesess are performed just like any other memory read, in the same way as you access any other object. Of course it'll be faster than accessing an SQL database.
The Prevayler people are very good at twisting the truth to create some amazing benchmark results. It's a good technology, but the the attitude of the developers is slightly irritating.
eMachines? What's that? Oh, never mind, google told me.
And by the way, how many GNOME applications can you think of that doesn't use Control-C etc, for cupt&paste (unless you configure it otherwise).
I never expected you to be. The intention was to convince other readers.I guess they weren't k3wl enough. You're not Internet-savvy if you have an "Office keyboard". If you got one of those "Web keyboards" though...
Before I installed RedHat 9 my GF was rebooting into windows all the time (I had this WindowMaker+assorted tools setup) becuase she just felt Windows was easier to use. After I reinstalled with RedHat 9 and decided to stop mess around I realised that she wasn't rebooting anymore. I didn't even ask her. She really felt it was nicer to use the GNOME desktop than Windows.
So, you can complain all you like about how crappy the Linux desktop is, but I have actual proof that at least one fairly computer ilitterate person prefers GNOME before Windows. (and no, she's no power user. she uses Mozilla, various text editors, Open Office, GIMP (for working with the digital photos) and I hardly ever have to help her).
Why is Windows so consistent in having Control-C etc... being cut&paste keys. They add a million "email-buttons" and other crap to the "windows keyboards" but they fail to add the thee most important buttons...
Cut, Copy and Paste buttons.
Sun has had them for 15 or so years. Fortunately I can cofigure my Sun keyboard I have connected to my Linux PC to use these buttons for copy&paste. I have yet to find a way to do it in Windows.
You are still allowed to use the same algoritm in different circumstances though. Read the article.
This isn't the first time seemingly good things has been announced which dissapointed later. I'll rejoice when I see some positive results.
Hah! It only looks better if it's GNU Emacs. XEmacs runs better with with ATI.
I honestly don't know where they provide this information. I had connected everything before they even installed the switches in the basement of out apartment building, and started using the stuff when I saw the lights on my switch activate. :-) As soon as I noticed they were blocking incoming traffic I opted out.
I still think that their way is the best. It will protect most network-illiterate users, and still open up everything for us who want it.
Oh, and by the way: Even before I opted out of their firewall, I could play pretty much all online games (but not host). So I suppose very few people will even notice they have the firewall.
I loved the Jeff Minter games. It's great to see something new coming out which is similar in strangeness.
What's the density of silicon?
Maybe it's just me... I've never had a single post being modded up as "funny". I should really stop trying.
The iRiver players (I'm a happy owner of the IMP-400) has at least 180 seconds of shake-buffer. I've never had a problem with shaking.
Thank god the packages are different or we would have millions of "consumers" playing The Matrix in their car CD-players.
For example, what will the value of "bar" be after the following code?
In Java today, I'm 100% sure that bar will be true. After 1.5 (and in C#) I don't know this, it can be true or false, depending on wether x and y are Integer's or int's. Integer and int are not the same thing, and it should be evident from looking at the code. That's the orthogonality I was referring to. Well, look at it this way:It's my firm belief that these features (especially static imports) were added for no other reason than to shut up the worst C# whiners.
The enhanced for loop has a horrible syntax, which only saves a few keystrokes at the expense of being very hard to read.
The autoboxing breaks othogonality in object identification. Consider the following:
(how do I ebped code in a post and retain the leading spaces?) C# programmers don't care about this very much because they already have this problem times 1000 because value objects and object references have the same syntax. Type-safe enums are just a compile-time code generator that creates the same kind of type-safe enums Java programmers have been using for ages. However, being able to use them in swich/case statements is an avantage. Static import is just plain silly. You save a few keystokes when calling static methods (something you do quite rarely anyway) at the expense of getting a lot more confusing code. The only situation where it might be nice to have is when you wring lots of mathematic expressions. I don't think this advantage outweighs the problems though. Metadata: why not simply extend the javadoc syntax?I'd be very happy to hear from someone who tried both, and still like VS.NET more.
ASCII, yes. But there are a few hundred or so thousand or so Unicode characters to choose from. They could be using them too.
Has anyone experienced this before?
Much of what you sait made sense back in the 90's. These days it's a completely different picture.