Slashdot Mirror

Here I give a *scientific* paper evaluating the performance of Java

Sigh. Here is what you did. You sited a paper, correct, that says, among other things "e first perform some micro benchmarks for various JVMs" (my emphasis), then you said "Only n00bs believe marketing spiels and microbenchmarks". See the problem?

Incorrect. Java is much older than C#

Sigh. If you don't understand the difference between "old" and "mature" then you need to stop conversing with adults. Yes, Java is (obviously) older than C#, but development in Java, that is, the development of the language it self, has been extremely slow, and Java has not incorporated newer and better language constructs at the pace C# has. Java was at the forefront of popularization of certain things like VMs etc, but that was long ago. Java has basically not evolved at all since Sun added (in a terrible way) Generics to the language. C# has evolved significantly faster, and now leaves Java in the dust. C# generics are done right, Java - wrong. Java auto boxing is basically a bug, not a feature. Java has nothing like the dynamic aspects of C#, functional programming is (basically) totally absent from Java. Java has nothing like the async support of C#. Hell, I can only say LINQ, and Java is instantly old, decrepit and from the 1990s.

If by more developed you mean more complicated and with a rapidly increasing number of constructs then C# is indeed ahead

I guess that is what adding modern programming constructs looks like to someone who is unable to learn.

Will third parties access your back end

I build enterprise apps. The answer to that is "always".

there is nothing to fear with a JSON interface so it is strange you do

Sigh. Why do you think I do? I specifically mention REST above. REST returns XML or JSON (or ATOM or...) based on what the client asks for. Assume you hava a method that returns a list of customers from either from an Oracle database (using EF, Hibernate or similar here). You need a REST API that returns that filtered by the start of the customer name (overly simplified here). The code (all that is needed) would look like this:
public void CustomersController() {
theDB = ... connect to the DB;
}

public List<Customer> Get( string nameStartsWith) {
return theDB.Customers.Where( cust => cust.Name.StartsWith( nameStartsWith )).ToList();
}

This code will respond to a request to http://host/api/Customers?nameStartsWith=John

As I said, this code will return JSON if the client asks for JSON, XML if the client asks for XML etc. It also shows some LINQ above. LINQ is excellent and a huge time saver. You can use an SQL-like syntax to querey anything. Whether the object "theDB" above was a Hibernate construct connecting you to an Oracle DB, or it was an in-memory XML stream read from a config file (for example for testing) or a List<Customer> or anything else that is queryable, the code is identical. I'd like to see you do this in Java. Seriously, I thought everybody knew that REST typically was JSON. Luckily I won't have to force my clients to chose, if they want ATOM it returns ATOM, if they want XML, it returns XML, if they want JSON it returns JSON.

make a Java webservice client

I don't have to chose. Once I build it on web api (also known as WCF), and I build it ONCE, the client decides what he wants to use for accessing the data. SOAP, XML serialization, JSON, ATOM, you name it.

Betting the farm on Microsoft and Microsoft tech is a strategy that people used to do a decade ago

You have never worked in an enterprise have you? AD everywhere. Exchange integration mandatory. That's the enterprise world of today.

Oh, I did not say that I trust Google. But so far, they have not been evil (excluding the china deal).

As to them having my data in cache, great. I say go for it. It means that it is one more way to reach my site. Likewise, they can use my content if it brings me more users. That is the name of the game, right? The funny thing is that most businesses and content developers would gladly give access to content if it brought them more customers.

So, you want to topple Google? Well, it is possible to do right now. The first thing is, that you need to get content. In addition, you need to direct traffic to your search engine. Of course, that will mean getting all the browsers to use you, not Google. But doing that would appear to be difficult, right? Wrong.

Here is something that I have been thinking of doing. Apache is the main webserver in http space (and I believe in https space). But one of the things that it lacks is a GOOD search engine. But you could easily create a module whose only purpose is to re-direct a search to some other search mechanism. The way that I have been thinking of doing this, is to allow URL like http://host/search. And allow for sub dirs to be append to this (http://host/search/data/abcd when you were originally in http://host/data/abcd). This module could call a local search engine or a remote one. By default, when apache is installed, the search mechanism would then look at a file and randomly pick a search engine. If the installer wants to override, they can. They can pick a different remote search engine or they could install a local one (useful for doing a DB set-up). So what advantage would this have? It would allow for == considerations by web-servers for sending data around. From this point, you need to have a good engine, but at least you would have a fighting chance. Without it, you are pissing in wind by trying to convince the end-users to switch. I have even thought of several other approachs to how to beat Google. It is Possible. Beating MS with their illegal monopoly is the tough one. That requires exactly what is occuring; Lots of work over a LONG period of time by the industry. No one company can take them down. Sadly, they will remain an evil company.

Oh, I did not say that I trust Google. But so far, they have not been evil (excluding the china deal).

As to them having my data in cache, great. I say go for it. It means that it is one more way to reach my site. Likewise, they can use my content if it brings me more users. That is the name of the game, right? The funny thing is that most businesses and content developers would gladly give access to content if it brought them more customers.

So, you want to topple Google? Well, it is possible to do right now. The first thing is, that you need to get content. In addition, you need to direct traffic to your search engine. Of course, that will mean getting all the browsers to use you, not Google. But doing that would appear to be difficult, right? Wrong.

Here is something that I have been thinking of doing. Apache is the main webserver in http space (and I believe in https space). But one of the things that it lacks is a GOOD search engine. But you could easily create a module whose only purpose is to re-direct a search to some other search mechanism. The way that I have been thinking of doing this, is to allow URL like http://host/search. And allow for sub dirs to be append to this (http://host/search/data/abcd when you were originally in http://host/data/abcd). This module could call a local search engine or a remote one. By default, when apache is installed, the search mechanism would then look at a file and randomly pick a search engine. If the installer wants to override, they can. They can pick a different remote search engine or they could install a local one (useful for doing a DB set-up). So what advantage would this have? It would allow for == considerations by web-servers for sending data around. From this point, you need to have a good engine, but at least you would have a fighting chance. Without it, you are pissing in wind by trying to convince the end-users to switch. I have even thought of several other approachs to how to beat Google. It is Possible. Beating MS with their illegal monopoly is the tough one. That requires exactly what is occuring; Lots of work over a LONG period of time by the industry. No one company can take them down. Sadly, they will remain an evil company.

Yes, I understand those things. And implementing user@host on top of a fixed mapping to http://host/openid/user or something would be fine. My point is one of usability - people understand and work with email addresses to represent a logical identity all the time. Why confuse the matter now simply for the sake of removing 5 lines from the spec?

JWZ is annoyed about the sound system in Linux. Well he is right about that. ALSA/OSS can have strange setups. And sometimes it refuses to work at all. But this only happens when using handmade kernels, distributions (gentoo) or certain releases of debian sarge and sid. On the other hand. I've installed Linux the past couple of years on various machines without a problem. I'vs used SuSE Linux, Knoppix, ubuntu and debian. All of these distributions had never a problem to detect the 3 button-wheelmouse. And even when changing the device with another mouse. Everything worked fine. The monitor setup was good (using the suse stuff it was perfect => better default than windowsXP/NT/98 on the same machine). The sound worked on all machines out of the box by auto detection (except from my desktop machine, which uses a non-PNP ISA SB16). So you can see, I cannot follow the argument. ALSA or OSS don't work. They might not work on all hardware. Or maybe some distributions are not that good in setting up ALSA correctly. But that makes not Linux (the kernel) a bad thing at all. Another point is: MacOS X work fine (almost). Except the network printing over ipp. It generates the worng URI for the printer. Because it assumes that the correct URI is: http://host:631/ipp/PRINTER-NAME or ipp://HOST/ipp/PRINTER-NAME Well, on most installations of cups. The correct path would be ipp://HOST/printers/PRINTER-NAME Even on other MacOS X machines. So this is a serious bug in Mac OS X. But you can fix it. And to be honest, iCal is really cool. And the other nifty things are cool too. e.g. iPhoto which is just easier to use then gthumb or the KDE equivialent. So I can see, that someone might just get a Mac so he or she can work a little more efficient than on Linux or Windows. No problem. But I really don't know why it is so important that jwz is stopped using Linux as desktop-OS. Is he some sort of prophet of the Linux-Religion? Or is he a fallen angle? Nope. He is just a man. And if he stopped using certain open source software. Well why should we care?

GWA also doesn't prefetch GET with query strings. The problem is that apparently Basecamp/Backpack uses short/pretty URLs that don't contain query strings, e.g. http://host/account/delete/121 instead of http://host/account?action=delete&id=121. It's not prohibited to use GET for delete/add/whatever links.

GWA also doesn't prefetch GET with query strings. The problem is that apparently Basecamp/Backpack uses short/pretty URLs that don't contain query strings, e.g. http://host/account/delete/121 instead of http://host/account?action=delete&id=121. It's not prohibited to use GET for delete/add/whatever links.