rlkBlog

MozCreator progress: 20050715

2005-07-15T14:08:00.000-07:00

Just fixed the MozCreator project tree. It looks really great! Now if I could just get the folders to open and shut :). Here's my to-do list for MozCreator-Eclipse:

- Fix properties table. Layout issues with it and re-hook it into the TodoTable class.
- Tweak the XulParser API a bit before releasing 3.0.
- Get properties dialogs working and fully functional.
- Status bar and source highlighting?

I'm planning on releasing a MozCreator 0.2.5 (yes, I'm going to version numbers) since the new Eclipse release doesn't feature anything new other than the move to Eclipse from Swing. This should also eliminate the localization issues people were seeing.

Picture:

MozCreator now builds on Eclipse!

2005-07-06T18:39:00.000-07:00

Screenshot here -> http://www.rj.pcblaze.com/mozcreator/screenshots/mozcreator-20050706.jpg.

Now that doesn't mean it functions under Eclipse. I gotta move some of the JFrames to Shells still, but a lot of progress has been made. There's also a HUGE performance improvement and the JAR file is smaller.

More to come...

XulParser 3 Progress: 20040628

2005-06-28T17:25:00.000-07:00

Current bugs:

Painting issue with tabs
Implementing wizards is harder than I thought. The JFace interface doesn't like my parser architecture, but shouldn't be a huge problem becuase I can just use XulWizard class that will provide a decent workaround by parsing the XUL when the dialog is first set visible

After that, it looks like we're home free! I'll checkin the code and we'll be up and runnin' :). Eclipse platform has a lot to offer. I'm very excited that we're moving. Maybe in the future we can make a MozCreator plugin for Eclipse.

XulParser 3 Progress: 20040625

2005-06-25T18:29:00.000-07:00

Wow! I can't believe how much easier it is to implement Eclipse in XulParser using SAX than DOM! Who would've thought! Good thing too because SAX is faster.

Here's the pic:

For this xul file:


<?xml version="1.0"?>
<window xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul"
  width="425" height="160" title="Find">
  <label value="Text"/>
  <button label="test"/>
</window>

Not bad for a couple days of work :).

Theory: AOL still going on with its plans with Netscape

2005-06-24T19:38:00.000-07:00

If I can make a theory, I think that AOL is still going on with its plans with Netscape, just a bit differently. I bet they'll eventually integrate Netscape with AOL once their competitors get more standards-compliant (it's bound to happen sometime) and try and profit off of Mozilla's work. I just noticed that there are a ton of places on the AOL website where they have a "Download Netscape now" link.

I'd find it interesting if AOL tries a jump because that'd mean that Firefox's market share would surge big time. It'd be a huge hit for IE and a big win for AOL because they could get a edge on MS with a technology they have control over without needing to spend any extra cash.

I find it very ironic how successful Mozilla has been in general since Netscape was taken out of the picture. Now I see way more components owned by non-mozilla developers and a product that is more user-friendly and popular. You'd just think that the enormous amount of funding Netscape put into Mozilla would've caused it to have similar success.

When it comes down to it, the amount of money you invest in an open-source project doesn't directly result in an increase in success or a better product.

XulParser 3 dev started

2005-06-24T17:52:00.000-07:00

Just started development on XulParser 3 spec and checked in some code based on it. Currently I'm dumping all XulParser 3 changes on the MOZCREATOR_XULPARSER_3_BRANCH cvs branch.

Some current plans:

Move to the SAX parser. Should offer a performance improvement.
Making the big jump to Eclipse SWT from JFC/Swing.
Lots of changes to make the API easier to integrate with other Java GUI toolkits (like qtjava, wx4j, etc.)
Major API changes include making GuiParser no longer a abstract class, total removal of Events code, so it's completely unintegratable. Removal of IParserSettings and replacing it with a TagData class that's more SAX friendly.

As with each new full point release of XulParser, we're breaking compatibility with the old API completely, which means removal of all deprecated classes, including those of XulParser 2.1.

Making your Java XML code faster

2005-06-18T13:01:00.000-07:00

I've been reading some articles on how to optmize XML performance in Java, and this is what I got out of it:

Writing XML files:

Use US-ASCII encoding. It's faster because characters are a single byte. UTF-8 will be just as fast on Xerces2 but only if you only use ASCII characters
Don't type a lot, like don't put a million code comments (duh!). A better idea might be to use something like the XUL preprocessor over your XML code.
Don't use namespaces. It's slower
Don't include doctypes unless you need to (like in the XUL Parser's case, you'll need a doctype if you use DTD files for localization, but should remove it if you don't). DTDs only make the parser do more work.

I'll probably integrate these tips into the MozCreator Coding Style Guide once I get it up and posted (it's currently floating on an internal wiki).

Rhino to ship with Mustang?

2005-06-18T12:38:00.000-07:00

Wow! This would be so awesome! Got it from a quote from this sun article.

"Aside from the framework, we will also include a JavaScript engine based on the Mozilla Rhino implementation."

Overall though, I'm disappointed that Mustang we didn't get more of the C# features like properties (PLEASEEEEEEEEEEE!!!!) and I really want operator overloading. You can say all you want, but operator overloading would be so sweet. It's obvious that it's a useful feature because Java does it itself with the String class. So why can't we use it with other classes?

Someone also needs to explain to me why Sun continues with their crap UI system. Swing sucks so much. I mean, it's a well-designed API but it's slow and looks horrible. Native GUI like Eclipse (or QtJava or wxWidgets4Java) seems so much more logical. Now you may be saying to me "well just use their native GUI platforms and stop whinning", bu their layout managers just suck. Why can't I get a BorderLayout? I'm sure if I research QtJava more (which looks like a much better GUI platform), then I'll be happy but I just don't get why sun has to continue with their old, native-simulated GUI.

Maybe one day we'll get a full, real open-source Java so people can start shipping their own distributions with some real features for once. At least I'll get Rhino :D.

UnspportedException

2005-06-18T12:25:00.000-07:00

This is a interesting blog post I read -> http://pezra.barelyenough.org/blog/2005/06/why-java-is-not-my-favorite-language-%e2%80%94-reason-16/

I see what he's saying, Java is really designed to be more object-oriented than logical in a lot of situations. Although I'd really hate for the remove() function to be split into their own separate interface. The collection interfaces are icky enough as it is (if only it was all like Map, it'd all be so much better).

MozCreator XulParser box model

2005-06-16T20:11:00.000-07:00

Just checked in a huge reworking of the way MozCreator parses XUL files and makes it MUCH more compliant with the Mozilla and is much easier to make your GUI look perfect. Basically uses , and flexes instead of the old jxul:layout="BorderLayout" technique that was really messy.

This combined with Rhino support has really made huge strides in compatibility with the Mozilla platform. It'll be interesting to see where we're at in a couple of months from now.

WindowsN

2005-06-08T15:17:00.000-07:00

So the new Windows without Windows Media Player is called Windows-N, which Microsoft said means "No", or to me, "Do Not Buy". I would so love a copy of this OS! I never use all that WMP crap. Small OS == good.

News.com article ->
http://news.com.com/Microsoft+updates+slimmed-down+Windows/2100-1016-5737415.html?part=dht&tag=ntop&tag=nl.e433

laid off ;(

2005-06-03T08:18:00.000-07:00

I was recently laid off of my job at TRF Inc. My expertise is in web design and Java programming but I also have experience working on Mozilla (although I possess little C/C++ skills). If you know anyone that might need a web designer or java programmer, tell them to give me a buzz :). I'll post my resume once I get some time.

grendel picks

2005-05-06T13:09:00.000-07:00

Checkout these sweet Grendel pics! Click on picture to see full version.

VB Petitions are stupid

2005-04-30T19:46:00.000-07:00

Somebody needs explain to me the point in having a petition like this. To me, it looks like they want Microsoft to support Visual Basic 6.0 compiled code in the .NET IDE.

Please tell me why you would want to sign this? If you like VB 6, then use VB 6. What's the problem? Even if Microsoft were to add VB6 support to Visual Studio .NET, would it be any more useful than the current VB6? Probably not.

If you like VB 6, then use VB 6, don't go signing a petition to add useless new features to a product. I'd hate to pay for VB 6 .NET with Visual Studio Professional.

Leave a comment if you disagree.

XUL Parser interview

2005-04-30T10:31:00.000-07:00

I was asked to do an interview on the MozCreator XUL Parser, so I thought I'd post the interview below.

Can you tell us a little bit about yourself?

My name is R.J. Keller and I work as web coordinator of TRF Environmental Inc. My primary expertise is in HTML, CSS, PHP, ASP, MySQL and Java programming (although I tend to do some C# and Visual Basic programming from time to time).

My main work on Mozilla includes being former module owner of the Help Systems module, the creation of the Firefox Help Viewer API, and coordinating development around it. After my resignation as module owner, I now do general patch work on Firefox and Mozilla Suite when I have time.

How did you get started on the MozCreator XulParser? Did you look into any existing open source XUL parsers in Java before starting work on the MozCreator XulParser? Can you tell us a little bit about the MozCreator XulParser's history?

The MozCreator XUL Parser came about when we wanted the entire GUI to be written in XUL but with the back-end code written in Java. We looked at several ways to possibly create a bridge between Mozilla and Java but few of them worked or were stable enough for our needs.

After that, we decided to look at XUL parsers written in Java. The issue with those was that they were difficult to use and didn't really suit the Java programmer well. We needed a compact and easy-to-use solution. Swing turned out to be a nice GUI platform to build a XUL parser on top of. We did run into some issues where some core Java GUI features weren't available, so we created a XUL add-on called JXUL to get some of those features available through out XUL interface. The language ended up being very nice to work with considering we had all of the Java language features in an easy-to-use XML format with little performance hit. It's also very easy to make a GUI class use XUL, by simply running the code "new XulFrame("C:\myFile.xul")", you have a JFrame object containing all of the GUI elements parsed by the XUL parser. Or if someone doesn't feel like moving all of their GUI code to XUL but just some of it (for example, just the menubar or toolbar), then you could use the main parser interface to parse only those elements.

Can you briefly sketch out XulParser's architecture and its building blocks?

For usage, you don’t need to know much. Basically XulFrame takes in a file name through the constructor plus an array of Event objects, which basically associates a string with an ActionListener so you can set “onclick” fields in your XUL file. The only other class you might need to use is XulParser, which contains functions for getting a JAXP XML Document object and using that to parse only certain tags of your document individually. This is useful if someone wants only the tag. Usually it’s a bit tough to use so using XulFrame will make your life easier.

Going a bit more deep down inside the parser, we contain abstract class GuiParser that can be used to implement any XML-based GUI language. XulParser is an implementation of this class. GuiParser parses tags based on objects of the ITag interface. ITag interface defines functions that, if you implement yourself, can return a component that implements JComponent to use in the parser plus make it easier for attribute and child tag parsing. Generally to add your own tag, it takes very little effort. Basically uniting this interface anyone could create their own XUL-based language with relatively little effort.

Can you tell us how you handle the mapping from XML tags/attributes to Swing classes/properties? Do you use reflection? Do you use hand-coded glue code?

Yes, we do use reflection in some places. Basically what happens is you implement the ITag interface. When you implement GuiParser, you provide it with an array of class objects for each of your ITags. These are used by the GuiParser to create new instances. Then once GuiParser creates the new instances, it runs the parse method which basically returns the JComponent. Additional information like the GuiParser interface and the parent and main window elements are provided through parameters to the parse method.

For attributes, you can extend the IAttributeHolder interface (which extends ITag). This provides an extra method called setAttribute that sends JAXP Attr objects so you can set the swing properties of your JComponents. Just using a simple if statement, it checks to see if the attr equals one of the valid attributes and if so, sets the swing element appropriately.

If your tag can contain other elements, you can extend the IContainerTag interface, which adds method “add”. This takes in a DOM Element which you can use the GuiParser instance given in the parse method to parse this element and add it to your container in about 1 or 2 lines of code depending on the container.

Generally the parser handles all the rest. Overall it’s easy to add your own tags to XUL if you wanted to or add support to a previously unsupported tag. What I like about this parser is that, if you run into about 10% of tags that are difficult to add support to in the parser, you can get around that problem easily with little performance hit. I’ve yet to run into a situation where a XUL tag couldn’t in any way fit this interface.

Can you tell us some challenges you faced creating XulParser?

Getting some of those strange tags to work. was extremely difficult to implement and adding support for the tag was difficult due to the fact that you couldn’t use XulFrame. The wizard problem was solved by creating a XulWizard class but it’s difficult to make custom window solutions.

At first, the interface was based on ITag only without the extra attribute and container interfaces. That caused for a ton of for loops and if statements. Now that is all handled through the new interfaces (IAttributeHolder and IContainerTag) and have saved us a tremendous amount of time.

The only main challenge we still see now with XulParser is having 100% perfect XUL be parsed in both Mozilla and XulParser. I’m hoping one day you could make a complex XUL file work both in Java Swing and Mozilla. It’s difficult to do that with swing elements, but is definitely possible.

Also a lack of JavaScript or CSS support has caused huge issues with getting styles and scripts to XUL tags. I do have a separate project I’m working on that integrates Rhino with XulParer. To fix the CSS problem, we’ve converted CSS attributes to XUL tag attributes, which is very messy but adds the functionality. I’m planning on adding a ICssHolder interface to add some basic CSS support to hopefully improve support for XUL in XulParser.

Can you tell us what the MozCreator XulParser can do today? What works and what needs to be done?

Currently, it can take a basic XUL file and run it in swing decently. MozCreator IDE is written entirely in XUL and runs good.

What needs to be done is CSS support. This should fix a ton of bugs in the parser and could make a Mozilla/Java cross-XUL platform realistic. We also need to improve our layout of XUL in general to be more precise with Mozilla’s layout.

A XulWriter is also in the works to be used with MozCreator’s GUI editor. We’ll have more on that at a later date.

Do you have a favorite scripting language for the Java runtime? Any plans for adding support for scripting to XulParser?

Yes, as mentioned above, we are planning on adding Rhino + XUL integration to make a full version of the Mozilla Platform available to Java users.

Any plans of supporting different UI toolkits such as SWT, wx4j or Java Gnome, for example? Or do you plan to stick to Swing?

Actually yes, recent modifications in the XulParser 2.0 spec make Eclipse SWT support possible. Progress is slow, however, due to a lack of contributors.

With the current parser, if someone wanted, they could easily port a Java GUI toolkit to XulParser. The issue is getting the manpower for it.

Can you tell us how popular the MozCreator XulParser is? (e.g. How many downloads? Are there any applications/projects using XulParser? What's the interest in the Mozilla or in the Java Swing community? etc.)

It’s difficult to track downloads because the project has been jumping FTPs often. I do know that the number of downloads is way higher than we’ve expected. Just seeing the surge in bandwidth usage on my FTP since MozCreator was added makes me confident that we’re getting a large number of downloads. Maybe in the future we might be able to give some numbers.

Who else is behind XulParser? Do you work on your own? How much time do you spend on XulParser development? How can someone get involved in the MozCreator XulParser project?

The team is small and we need contributors badly! Currently I’m the only programmer but Brant Gurganus and Ryan Probosco along with others have made contributions and ideas to the specification.

Currently I spend about 3 to 4 hours a week on MozCreator and the XUL Parser.

Does the XUL handled by the MozCreator XulParser differ from Mozilla's "classic" XUL? Do you envision XulParser to run Mozilla XUL apps out-of-the-box?

JXUL is the main difference. We definitely hope to have any Mozilla XUL file run out of the box on MozCreator.

What's next for the MozCreator XulParser?

As for the parser itself, better layout and easier GUI parser support.

finally, some XUL gui editor

2005-04-29T19:53:00.000-07:00

Just posted a nightly build of MozCreator with the new XUL GUI editor. Currently it supports drag-and-drop GUI editing by using a stack. Once I get that fully working, I'll move on to working with the box layout more.

Don't expect too much. It's only a prototype and isn't even close to alpha quality. You have about a 90% chance that it'll throw an exception somewhere when trying it out but might give you a feel for what's coming up.

Some tips:

Click on a control and then the form to add a control.
Double click a control (any of them) to see the generated XUL code.
If the control doesn't have an icon, it probably won't work.

Hopefully will have some bug tracking on the GUI parser set up soon.

Also, this is the smallest MozCreator yet! Over a full meg smaller than the previous nightlies.

Back Home :)

2005-04-27T19:16:00.000-07:00

Just came back from BPA in Anaheim, California. Won 1st place in Web Development Team. Tons of fun. Currently getting catched up on stuff like email.

While at California, I made a GUI editor for MozCreator. It generates XUL and has properties that change. A bit buggy at the moment but not bad for a prototype? Hopefully we'll have more to see soon. Most new stuff isn't on CVS yet. Hopefully school will lighten up a bit and give me some time to post some updates :). Did post a mini-MozCreator update over at MozillaWiki.

UltraSwing GUI

2005-04-03T13:29:00.000-07:00

Found this really neat GUI library called UltraSwing. Basically it adds several popular application features in major Microsoft products to Java Swing. I'm planning on integrating the Visual Studio .NET-like GUI into MozCreator which should increase its usability a lot.

So for all you people trying to learn the Eclipse framework to make an IDE, you might want to check this out. From my experience, Eclipse is overkill, even for an IDE. I'd prefer a simplier IDE like Visual Studio to do my Mozilla programming in.

Inefficiencies in Java Enums

2005-03-07T15:15:00.000-08:00

So Java 1.5 has this new feature called enums. The goal is to get around the problem of using primitive data types to represent a series of constants which is not type safe. Although Java enums sounds all fine and dandy, they are secretly converted into extremely inefficient code when used in certain situations.

Take, for example, a sample Java 1.5 enum example:

public class enumTest {
 public static enum rlkVars { TEST, TEST2, TEST3 };
 
 public enumTest(rlkVars var) {
  switch (var) {
  case TEST:
   System.out.println("test");
  case TEST2:
   System.out.println("test2");
  case TEST3:
   System.out.println("test3");
  }
 }
}

In line 2 of this code segment, we have a simple enum statement. Since enums are technically considered a class, they are converted into an actual class by the Java compiler. Lets observe what the compiler converts that enum into(1).

    public static final class rlkVars extends Enum
    {

        public static final rlkVars[] values()
        {
            return (rlkVars[])$VALUES.clone();
        }

        public static rlkVars valueOf(String s)
        {
            return (rlkVars)Enum.valueOf(enumTest$rlkVars, s);
        }

        public static final rlkVars TEST;
        public static final rlkVars TEST2;
        public static final rlkVars TEST3;
        private static final rlkVars $VALUES[];

        static 
        {
            TEST = new rlkVars("TEST", 0);
            TEST2 = new rlkVars("TEST2", 1);
            TEST3 = new rlkVars("TEST3", 2);
            $VALUES = (new rlkVars[] {
                TEST, TEST2, TEST3
            });
        }

        private rlkVars(String s, int i)
        {
            super(s, i);
        }
    }

In this very large class, we see a number of inefficient operations taking place. The major operation is the creation of a new object instance for each enum value. Based on Java performance(2), the creation of an object takes over 653 times longer to perform for each object than the assignment of one integer constant. So basically, using an enumeration takes 653x times longer (with x being the number of enum values in your enum class) to just initialize your class than if you used regular integer constants. Is using enums worth the performance hit?

Sadly, it doesn’t get any better. Let’s look at our original code example at the switch statement in the constructor. This is a very basic operation to do with an enum (if you decide against a series of if statement). Anyone notice anything funny when you compile this code in Java 1.5? 3 class files are outputted instead of 2. Where is the third class coming from? When dealing with enums, Java creates a separate class to deal with the switch statement. That class is shown below:

        static class _cls1
        {

            static final int $SwitchMap$enumTest$rlkVars[];

            static 
            {
                $SwitchMap$enumTest$rlkVars = new int[rlkVars.values().length];
                try
                {
                    $SwitchMap$enumTest$rlkVars[rlkVars.TEST.ordinal()] = 1;
                }
                catch(NoSuchFieldError nosuchfielderror) { }
                try
                {
                    $SwitchMap$enumTest$rlkVars[rlkVars.TEST2.ordinal()] = 2;
                }
                catch(NoSuchFieldError nosuchfielderror1) { }
                try
                {
                    $SwitchMap$enumTest$rlkVars[rlkVars.TEST3.ordinal()] = 3;
                }
                catch(NoSuchFieldError nosuchfielderror2) { }
            }
        }

        switch(_cls1..SwitchMap.enumTest.rlkVars[rlkvars.ordinal()])
        {
        case 1: // '\001'
            System.out.println("test");
            // fall through

        case 2: // '\002'
            System.out.println("test2");
            // fall through

        case 3: // '\003'
            System.out.println("test3");
            // fall through

        default:
            return;
        }

This class is initialized to return a value for the proper enum in order to feed to the switch statement and use. Notice that if we had a worst case scenario for the switch case, rlkVars.TEST3 was the switch value, we would go through a total of 3 try-catch statements, 2 of which would throw exceptions (which take 320 times longer per exception thrown than an integer comparison). Then, once this is all over, it makes a normal integer comparison using the switch statement that would’ve been done if we had only using integer constants. So in a worst case scenario, the switch statement would take 320x times longer to perform (with x being the index number in the enum) than using our switch statement with integer constants.

So overall, to complete this class in worst case, it would take 2,919 times longer to perform this very basic operation using enums than integers. Of course, normal integer constants are already super-fast, so 2,919 times longer would mean it MIGHT hit a full second to execute on a decent system. This isn’t so bad considering the enhanced program readability, but what if you had a 5000 element enum? This kind of operation would be insane to perform (not to mention a lot of typing :)). Maybe we should wait until Sun optimizes Java better before moving all of our integer constants to enums.

(1)All Java compiler conversions are observed using the JAD Java class decompiler.
(2)Performance diagrams from Thinking in Java 1st edition “Performance”, Bruce Eckel, Prentice Hall. Book available for download.

mozilla.org email addresses

2005-03-06T15:26:00.000-08:00

I read up on Mitchell Baker's proposal on staff members and who gets mozilla.org emails. I'd like to comment on the email portion specifically. If I was in charge of a policy, I'd like to see something happen like NetBeans, where people with CVS access get one @mozilla.org email forwarder. Email forwarders show pride and envolvement in a group. I personally would love a mozilla.org email just to show that I work on that project and to show that I'm not just a everyday contributor.

I think that it'd be a small and nice reward for those that do extensive contributions to mozilla. Staff emails should be @mozillafoundation.org to show that they are above the normal mozilla.org emails.

ahh, the good old days :)

2005-01-20T17:14:00.000-08:00

Anyone remember Mozilla classic? I thought I'd post some of my old Mozilla Classic builds I found on my old Hard drive when cleaning it up. They're quite interesting in seeing how far Mozilla has come, and a nice piece of history. Also featured if the first version of Mozilla compiled when the source was first released. Go check it out -> http://www.rj.pcblaze.com/archive/Netscape5/

Seperation of Church and State: My view

2005-01-16T14:19:00.000-08:00

Wow, some blog entries seriously disturb me. One of them is by Jacob Steenhagen about why evolution is taught in schools without other views and why the ten commandments.

Personally, I would not like the ten commandments to be seen in my courtroom. I won't say that I don't believe in God (I am a former Christian), but I would feel like one religion would be imposing their views upon me by having that displayed. Of course Christans would disagree with what I'm saying, but I'd love to hear someone atheist to want the ten commandments on the court room walls (and it won't happen).

And on evolution, why not teach it? I don't get how someone can deny evolution. A good book to read on this topic is The Way of the World (first part of the book) by David Fromkin. He gives me a good theory integrating religion with evolution that seems much more believable than anything else I've seen. If someone is denying that evolution exists, then they must be crazy. Genetic deformities happen all the time and survival of the fittest will ensure that someone who has a deformity that will give them a physical advantage over other beings will become superior. I can understand if you don’t believe in the whole money thing, because really I can’t see that entirely, but to say that evolution does not exist in any form is like saying that outer space is fake.

Why does Java have to be so stupid?

2004-12-29T14:08:00.000-08:00

Try and guess (and don't cheat!) what compile error this code will give:

JInternalFrame myFrame = new JInternalFrame();

myFrame.setSelected(true);

Want to know the answer? It complains the following:
unreported exception java.beans.PropertyVetoException; must be caught or declared to be thrown

What the hell? PropertyVetoExcpetion? What the heck is that? Is my property going to be vetoed :). Maybe I should have all of my properties in all of my classes decide to throw PropertyVetoExceptions in case I decide that they are ugly (not obsolete becuase then I'd put a deprecation mark on it). Can somebody please explain to me the logic behind this?

What's even better is the javadoc info on why the exception can be thrown.
PropertyVetoException when the attempt to set the property is vetoed by the JInternalFrame

So when exactly would the property be vetoed by the JInternalFrame? Maybe when Java considers my code too complex and decides to find a reason to bomb it.

Merry Christmas everybody!

2004-12-25T10:29:00.000-08:00

May everybody have a very merry Christmas!

Java XUL Parser?

2004-12-22T17:11:00.000-08:00

Yup, I got a XUL Parser written in Java working. HUGE improvement over the current Java GUI engine and has really helped me make advanced Java GUI solutions.

I've decided to restart the MozCreator project and will be checked in the parser files plus a acutal working IDE to the site. Currently CVS has been totally cleaned up. Not even previous revisions are there. I'll check things in once things stabilize.

MozCreator is really a nice IDE. I don't think that I even realized what awesome code I put up there. Hopefully I'll have some nightly builds for people to try out. Might even make a project file for Nvu Help to help with development there. Before releasing nightlies I am going to make sure the project files are written in XML files instead of object streams. This should allow someone to use MozCreator for a Mozilla development project realistically.

Nvu and Help

2004-12-03T18:58:00.000-08:00

Finally got some work done on Nvu Help and it's looking pretty good. Thanks to KDS for his excellent work on the documentation. Screenshot is here.

On another note, Grendel is starting to look really nice. It's really a awesome app. Go checkout the screenshot.

Grendel is alive!

2004-11-28T09:57:00.000-08:00

After spending about a week getting it to compile, I finally managed to get Grendel to compile and run today! What a monster app! Really contained some cool features. Should be interesting to see what I can implement in it.

In the changes, I removed about 6 classes and rewrote the XML Menu builder class to use my more XUL-compliant menu builder class that I wrote. Maybe in a couple of weeks I can restore functionality completely and get some of this stuff checked in.

Nvu is getting the heat

2004-11-28T07:10:00.000-08:00

After seeing a translated version of http://nvufr.mozdev.org/, I have to say that these localizers are really being idiots. Nvu is very open and does accept contributions. Sounds like they're complaining that Nvu is in development and not being frozen for localizations. And I say deal with it. It's too early in the process for localizations.

And for Daniel, just realize that, from my experience, localizers will never understand the development process. I was happy to leave Help Systems so I don't have to deal with them anymore :). So good luck daniel! More heat will probably come with time. I'd just get used to it because you'll always have people like that.

Netscape Prototype probably isn't based on Firefox

2004-11-20T13:42:00.000-08:00

Why would I say that? Just look at the screenshot, there's a toggle sidebar button. Such a button doesn't exist in Firefox. A lot of people in the Netscape forum are making the assumption that our NS Prototype is on Firefox when it probably isn't. If AOL is making this radical of a theme change, I wouldn't be surprised if there are truly unique features added to NS. I personally am a bit excited and curious to see what AOL has cooked up.

What I hope hasn't happened is AOL deciding to release NS using the IE rendering engine instead of Gecko. With this radical of a theme change, I wouldn't be surprised if they made an app not based on Mozilla (just using a somewhat similar look). Just think the AOL Browser. It's more than possible that they are going to be very similar in the coding. Just think AOL Communicator.

Netscape == Firefox?

2004-11-18T14:42:00.000-08:00

CNET has an article on Netscape that I find a bit disturbing. It sounds like AOL is moving from Seamonkey to Firefox and taking credit for all the new features. I know that the AOL spokesperson says that it's beyond Mozilla, but is it really? I interpret that as beyond Seamonkey. It'll be interesting to see what AOL decides to do.

Update: I really did mean to put a question mark in the heading :). Sorry for any confusion this may have caused.

sick

2004-11-16T10:36:00.000-08:00

I've had food poisoning for the last two days and just started recovering today. If you sent me an email, I may not get around to reading it for another couple of days (homework from school slows things down :)). Sorry for being so slow.

Java needs a new text API

2004-11-13T19:05:00.000-08:00

I decided, being bored on a saturday afternoon, to create a WYSIWYG HTML editor using JEditorPane and the HTMLEditorKit API that comes with Java 1.5. You'd think it'd be easy. Just a matter of saying "when the bold button is hit, make the text bold!". Oh wait, no. This is JEditorPane, so instead it will only make selected text bold. So what happens when no text is selected? It forgets the formatting.

After spending the past 3 hours searching newsgroups to find a way around this (as well trying hacks of my own) ended up in finding nothing. What bugs me the most is that the answer is right in front of me. Java comes with a Stylepad Demo that does the effect I want. I can't find the Bold code anywhere. It must really be burried because it seriously isn't anywhere to be found!

Maybe I should just make my own API or try and inherit StylePad. Only problem is that then you inherit the toolbars and all that other junk, plus not figuring it out for myself isn't cool.

If anyone knows a solution, just post a comment :). If I do decide to rewrite the API (so it's at least somewhat usable), then I'll probably post it with the MacroToolkit.

Microsoft says Firefox is not a threat?

2004-11-11T19:06:00.000-08:00

If Microsoft is really saying that Firefox is not a threat, then they are really stupid. Any competitor can be a threat. Saying that Firefox is not a threat to IE is being stupid and egotistical about your product. I personally think that Microsoft is taking Firefox way too litely. One day I have a feeling that they'll realize that their market share is diminishing and there's nothing they can do about it.

Too bad Microsoft wasn't charged with being a monopoly. If IE wasn't shipped with Windows, I doubt they'd be this confident.

Firefox 1.0 coming!!!!

2004-11-08T15:29:00.000-08:00

According to mozilla.org, Firefox 1.0 is supposed to be coming soon! How exciting! It'll be interesting to see how Mozilla's market share will increase after the 1.0 release. I would be surprised if it didn't take a huge surge. You'd be surprised how many users won't install a product until it's a final release. If Firefox got that much response being only in Beta, the final will be insane!! I'm looking forward to it :).

Kerry-Edwards 2004

2004-11-01T13:31:00.000-08:00

And yes, I did vote at MozillaZine that I'd rather have my presidental candidate win instead of Firefox succeeding. Yes, there are things more important than web browsers :).
&lt;a&gt;John Kerry for President&lt;/a&gt;

What could be so wrong with CSS?

2004-10-27T18:35:00.000-07:00

Maybe I just don't get it (which does happen sometimes :)). I currently contribute to the Neuqua Valley High School Web site. Recently I redesigned their Music page using brand new CSS instead of the HTML branding and cleaned up the content significantly.

Of course when it came time for approval by the webmaster (jeanne_heydecker[at]ipsd.org), it got scrapped. She didn't like the CSS. But why? Instead, she decided to remove my CSS in favor of font tags (probably the worst thing you could do IMO to a site), which is breaking lots of pages. Maybe I'm just stupid, but how could CSS be breaking consistency with the site when it generates the same style? And how does it decrease site managability?

I konw that I'm messing with the people above me, but if there's a better way to do things then why not do it? What bugs me the most is that she uses CSS for some of her ASP code. It's like as if she doesn't like me using my code but yet shes fine with CSS as long as she writes it. You'd think that 20+ years experience would allow someone to recognize the issues with a table and frontpage-based layout and want to move to something better. I did all the work, not like she has to rewrite it (because then I'd understand). And I have direct write access to the server, so its not like she'd even have to lift a finger.

I need to stop working on these crap sites and go work on some sites that I can be proud of.

Rot

2004-10-24T14:11:00.000-07:00

It’s a bit sad to me sometimes seeing that my former-module will never become what I envisioned it. In its current state, it looks impossible for any major innovation to occur. Got my Help 2.0 specification lying on my desk in front of me. Haven't yet been able to throw it in the garbage.

Sometimes it still saddens me, but in the end it happened because I chose it. Maybe working on Nvu help might satisfy my pains.

Thinking about starting a project on mozdev called help2, whose goal is to rewrite the Mozilla Help Viewer API according to my Help 2 spec. Basically rewriting the API around HelpDocBook and including features like Help Navigation toolbar, ability to add/remove sidebar tabs, and automatically generated TOC and Index. Stay tunned.

On other notes, I'm running with Beonex now, so give me a buzz, rj.keller [at] beonex.com, and am starting to go more into web design.

And if you want a gmail account invite (got 6), email me (rjkeller [at] gmail.com or beonex address above).

MacroTracker project will be restarted (again)

2004-09-20T17:31:00.000-07:00

I'm hoping to get the MacroTracker Project up and running again. More to come. I'll be posting a announcement and maybe even a 0.1 release by the end of the week :).

As always, I can really use some help. There's some good code there, and its pretty simple to learn. If you're interested in contributing, email me!

And unlike most of my Mozilla projects, I will be keeping this one to myself, mainly because its special to me (it was my AP Java final) and has been already really useful to me.

And if anybody knows, I'm trying to find a way to make a J# port of MacroTracker but be able to merge it with the current codebase. Sounds crazy? Should be easy if all I port is the UI. I'd like to have the XP of Java, but have it be fast on windows (hey, I run on windows and so does the rest of the world). And I got a XML reader working in Java/J# so I'll be merging that into MacroTracker for XP database creation (with J# and Java versions).

J# is really exciting technology (especially the Express version), and I highly recommend everyone to try it if they haven't yet. It really is an excellent language. Microsoft has done a excellent job with it.

The next revolution: HelpDocBook

2004-09-11T18:16:00.000-07:00

Now that I'm not owner anymore, I can concentrate on some cooler things :). One of them is evolving the Help DocBook API and its integration into Help Systems. Might start with Thunderbird Help. Currently I'm hoping to toss out RDF and XHTML and go with a nice clean DocBook format with content automatically generated.

People might not like this approach, however, because it's not as flexible as the old layout. On the flip side, someone could make their own XSL stylesheet to use, as long as they follow the Help Docbook format. Heck, they could probably make their own tags. Just might require some extra effort.

The goal is to make life easier for the localizers. My latest XSL stylesheet + example document is attached to bug 252180. Probably will be first loaded on MozDev.

what? overlays in HTML?

2004-09-06T17:53:00.000-07:00

Maybe I just don't get it, but why would we want to have overlays in HTML? For me, I'd prefer just having a nice, old-fashined #include-equivalent in HTML. This overlay sounds like crap and will only complicate HTML coding.

Its not like you can make add-ons to your HTML. There aren't any web page add-ons. It'd be pointless. Maybe the idea is to make a Web site more like a application.

I'm surprised nobody has thought of putting a #include-equivalent for HTML (and I don't mean stupid iframes because those don't work like a actual include). Its probably my #1 feature that would be VERY useful to me.

But in the end, all of these features are pretty pointless to implement. For me, as a web designer, I won't be using any of them unless IE supports it. Maybe Mozilla should think of just following Microsoft with its Web standards as opposed to making its own non-standard technologies (that suffer from a lack of interoperability). It'd definetely make Mozilla more popular on the Web rendering side.

gmail account giveaway!

2004-08-30T15:23:00.000-07:00

got 6 invites. mail me if you want one -> rlk [at] mozdev.org. First-come first-serve. If you sent me a request and it got lost in the large list of invites (either because of my filter or some other reason), mail me again for an invite.

darn makefiles

2004-08-29T12:45:00.000-07:00

maybe I'm just dumb, but I seriously cannot get my C++ app to compile using Makefiles when the CPP files are in a folder other than the root. It just can't travel into diffferent dirs.

I thought about not linking the files and just throw them in some dir. Oh wait, one problem, there is no cl parameter to change the dir!!!! grrrr!! If someone smart knows the answer, mail me or post a comment. My ultra-basic makefile is below and yes, I am using the Microsoft VC++ .NET 2003 compiler (hey, I like my garbage collector!):

build::

	cl /O2 /clr $(FILES)

Now that I resigned from Help Systems, topics will be less-mozilla related and more geek-related :).

update: got it working! Actually, its a bit of a hack but it works good enough :).

Removal of alternative stylesheets?

2004-08-24T07:56:00.000-07:00

Usually I support this whole idea of a UI Czar, but this is getting a bit crazy. Recently, Blake Ross fixed bug 253722 which removes the UI for alternative stylesheets from Firefox.

My issue with this removal is that it's part of the CSS2 specification. This is basically like removing standards compliance (usually a MS thing). From the way Asa makes it sound, it was decided at some aviary meeting that I never knew existed (probably for a reason).

I'm getting the impression that mozilla.org is thinking like comment 26 of bug 253722, lets just remove CSS2 support because it's buggy.

I understand that we're making a end-user product, but there isn't a lot of code with the Stylesheet switcher. It's not like View Source where we're removing a nice chunk of bloat. With all of these UI-removals happening lately in Firefox, I'm sure there is some decision that will annoy SOMEBODY.

Now I'm starting to support the foundation's continued support for Seamonkey, and may switch to Seamonkey myself. It has the features I need in a web browser and not just a window with a back button. Its sad that Firefox has been downgraded right before 1.0, but I won't rant any more about it because its out of my hands. Maybe a better task might be to compile the Firefox toolkit into Seamonkey, and then be able to use some of the features like Toolbar Customization in seamonkey. If it was this easy, it'd probably already be done by now so it must be harder than that.

I'm surprised they never proposed removing the Help Viewer.

IE support coming to blog

2004-08-12T12:46:00.000-07:00

For the first time, I viewed my blog in IE and realized that you have to scroll to the left in order to see all the text. I'll try and see if I can fix it. The Blogger ads are goofing things up.

But you guys should be using Firefox anyways, so go download it if you haven't already.

Nvu 0.4 has been released.

2004-08-10T16:38:00.000-07:00

ahh! Nvu 0.4! Our awesome Dreamweaver/FrontPage alternative. Now that this version is out, I'll try and cook up a Help Viewer using the documentation that KSS generously donated to the Help Viewer project and send it to daniel for 0.5. All depends on how things go. Won't make any promises.

Go grab Nvu 0.4

UPDATE: Nvu 0.4.1 was released shortly after 0.4. It fixes a problem with publishing to a web site.

The de-evolution of IE

2004-08-10T16:31:00.000-07:00

CNET has a really excellent article on the lack of evolution in IE. I quoted some portions of the article and my response to each below.

> a Microsoft spokesman said. "The current plan is to make new IE
> features available with major Windows releases...Aligning IE updates
> more closely with Windows releases benefits customers by minimizing
> the number of updates to deploy and service."

How is minimizing the amount of updates to the benefit of the consumer? That's like saying, "it's to the benefit of consumers to not add features that will help them".

A even more interesting question is how is this not a monopoly? It's not to the advantage of the consumer to get a free product when buying another one. Everyone uses IE and if they have to spend $199 for each Windows upgrade to get the latest version, it's definitely not going to go well with consumers.

This topic by itself could lead to the downfall of IE.

> Microsoft has countered its critics by saying the browser is not a
> stand-alone piece of software any longer and that there are substantial
> innovations happening across the Windows operating system,
> including browser functionality that is not immediately obvious to the
> end user.

If it's not obvious to the end-user, why are you implementing it? I know the word "immediately" is there, but still. I want to see features that I WANT and SEE. Not features that "you won't notice until you use the product for a long time". For the past 3 years, there has been no evolution of web technologies because of Internet Explorer. It'd time that Microsoft gets their act together and actually do something to the benefit of the technology community and its customers.

> The accelerated pace of security flaws in the browser has led some
> developers--and even high-profile organizations such as the U.S.
> Computer Emergency Readiness Team--to ask consumers to consider
> using a different browser.

Another case where Microsoft is screwing people over. More specifically, this happens with the IT field. I don't appreciate the fact that I have to install IE updates on all my company's computers despite the fact they don't use IE. I'd hate to be a person in a situation where they have over 700 computers to update that run IE.

Imagine if this happened with our government? Obviously they wouldn't be able to keep up with the updates on so many computers. Foreign countries could easily get a hack into our systems and steal our government secrets.

It just reminds me seeing that IT lady for my school going around on the computers in the CS lab and installing updates. It must be horrible thinking of how many more computers she'll have to do that on.

> Analysts say Microsoft has little incentive to release a new version of a
> product it gives away for free, in any case.
>
> "Internet Explorer is not a strategic technology for Microsoft,"
> Directions on Microsoft analyst Matt Rosoff said. "They would much
> rather have corporations use something like Outlook or Office to
> access information."

This is probably the best explanation for why standalone IE has been discontinued. If it can get consumers to upgrade to longhorn for the new IE, then it makes money. If it makes a standalone version, people will stick with the lower version. I've seen numerous times where companies upgrade to Windows 2000, skipping XP Professional entirely. Microsoft realizes that it needs this eliminated. IE is the way they can do that.

The sad thing is that it isn't to the IE users advantage.

Of course, we live in a society that uses the worst product. Most users use AOL despite the fact that it's constantly been rated the worst dial-up ISP. Many people use Pentium 4's despite the fact that Athlons are now far superior. IE is another example. Firefox has way more features, yet people use IE. What's even better is that you can add any feature you want to Firefox so that it suits your needs. Hopefully people will realize the advantages to upgrading to Firefox, especially after seeing the document.all support.

the NEW Java build system (sorta :))

2004-08-09T11:57:00.000-07:00

When running some tests for fun on my own, I noticed that every time I was calling a debugging function in Java, the size of the file increased. Not only that, when calling a debug function, a String object was created in some instances where I used substr() and decreased program speed.

So, while working on Mozilla, I saw the awesome XUL preprocessor. This is a C preprocessor that Hixie wrote in Perl. So I thought, wouldn't it be awesome if we could get this working with Java code?

Another thing was the make-jars.pl script, which could preprocess files and output them to a folder in a zip file. This is a very useful ability.

So what I made for MozCreator is integration with make-jars.pl and preprocessor.pl so that we can set environment variables when we want debugging code. When used wisely, we can also add/remove beta code or version information (which is faster than using a constant).

How the build system works is that there is a jar.mn file. The file is there to promote directory independence when coding but when compiled restores the default package directory. So here is an example of a MozCreator jar.mn entry:

*	org/mozilla/MozCreator/compiler/Compilation.java                   (compiler/Compilation.java)

The "org/mozilla/MozCreator/etc." part is the package directory. Where you want the file outputted to when compiled. Then the part in parenthesis "()" is the literal location of the file. The * at the start of the statement tells the perl script to use the preprocessor on the file before outputting. Then, all you have to do is execute a makefile command ("make -f client.mk build") and your files are outputted to the dist directory freshly preprocessed if you requested it.

So how does this work? The original make-files.pl file outputs a zip file. In order to get around that, we unzipped the file in the dist directory. Yeah, its dirty, but hey we're building!

The makefile also unzips all the classpath jars so that we don't have to set a classpath on the ant script in the dist directory. It makes things easier in the end, even though it sounds weird.

The greatest advantage of this system is its flexibility. You can insert almost an unlimited amount of debugging classes or precondition checking and not have to worry about taking a blow to user performance. It's adding one of C++'s excellent features to Java.

On the down side, line numbers are thrown off when exceptions are thrown. If you're smart with the preprocessor, this shouldn't an issue, but there are cases where it could possibly complicate debugging.

MozCreator is to a huge advantage by this system because it already requires it for making the XUL and JS files. Now we can add Java preprocessing without adding significant bloat to the build system.

Until I check them in on the MozCreator CVS server, the files are available at http://www.trfenv.com/rlk/MozCreator2.zip. I hope someone finds this useful :). I know a lot of C++ fans would. The manual to the preprocessor is at Hixie's site.

We need to thank mozilla.org for making all these excellent perl files available to us.

new MozCreator tab editor

2004-08-04T10:28:00.000-07:00

Below is a screenshot of me talking the NetBeans tabs and integrating them with MozCreator. At the same time, I added a JDesktopPane so that we can have one window for each editor.

I would post it, but it's extremely buggy at the moment. I'll checkin the changes once it's usable.

no more gmail accounts :(

2004-08-02T10:11:00.000-07:00

Sorry guys, but I'm out of gmail accounts. Feel free to send me an email if you'd like to wait in line. Google hasn't given me any invites for awhile, so you might have to wait.

another excellent Mozilla Platform language

2004-08-02T10:04:00.000-07:00

I was looking at the JRex Project at MozDev. I'm surprised mozilla.org hasn't looked at this project more closely. What it provides is a way for trusted JavaScript and Java to communicate. This is really an exciting addition to the Mozilla platform, and definetely much more appropraite to compete with .NET.

One of the big issues I see with the Mozilla Platform is the learning curve, but with JRex, all you need is a little bit of Java and XUL knowledge to make fast applications. With MozCreator, I'm planning on making it a bit like Visual Studio, where you can create a Java Mozilla Application, JavaScript, or maybe Python (since Brendan Eich notes its future support).

Its great to see that the Mozilla platform, using a bunch of already developed technologies, can compete with .NET easily. Mozilla already has some awesome networking libraries, now with the power of Java, you can do almost anything.

Since the Mozilla platform is a lot like .NET, you could combine code from all these langauges into one platform. Someone could code in whatever language they knew, while the other developers write in a totally different langauge. It's great to see variety and it can only take Mozilla to great lengths in the future.

Debian - the New Linux

2004-07-28T10:27:00.000-07:00

I just got a copy of Debian Linux on my desktop. After using Mandrake for so long, Debian is like a fresh breath of air. I never imagined that I could get all the features of Mandrake Linux for free!

Debian is a bit harder to use than Mandrake in terms that it doesn't have the "Mandrake Control Center" which greatly simplified system configuration. The big advantage of Debian is how easy it is to install software. Debian comes with just about every package you could want on a linux system and has a easy installation mechanism that automatically downloads the dependencies of software packages when they are not met.

Currently, the Stable Debian version is very obsolete, but the nightly build is new, stable, and features the latest packages. I highly recommend it as a Linux distribution for those more familiar with Linux.

Now, only one problem :). The bootloader didn't detect my Windows XP installation (and still doesn't :)), but did recognize my Windows 98 installation. I also can't find a way to access the information on my WinXP partition from within Debian. So, I'm currently without all my files :). If anyone knows a solution to this problem, please post a comment.

document.all has come!!

2004-07-26T10:54:00.000-07:00

Finally! As I said in a previous blog post, this is one of the steps necessary for Mozilla's success. The main reason why people don't switch to Mozilla is because their favorite sites don't work.

A lot of people have been complaining to bug 248549, but you need to remember that standards aren't quite what they used to be. In the past, the W3C used to be the dictator on web standards. Now, we have the WHAT WG to take web standards to the next level. Although the w3C is great, sometimes we need to make our own standards to go ahead.

With this addition of document.all support, Mozilla now insures interoperability with Internet Explorer and can finally bring some IE-only websites to Linux and other OSes without a IE browser. This decision was made for the good of the web, and I believe the Brendan and Johnny made the right decision. Users don't care about standards, they care about interoperability, and we're making an end-user product.

This isn't the first time Mozilla has sacrificed standards for interoperability. The marquee tag is a good example. We live in a IE world, and sometimes you need to make sacrifices in order to survive. Microsoft can use its monopoly power again Mozilla. In order to win, we need to play by Microsoft's rules sometimes.

Gmail account

2004-07-24T15:34:00.000-07:00

I got 1 (one) whole gmail invite to give away. More will probably be on the way. The invites are given out on a first come first service basis. Since not many people read my blog, the chances of you getting the account are pretty high.

Just send your first name, last name, and email to rjkeller@gmail.com.

vacation is over :(

2004-07-22T18:40:00.000-07:00

Just came back from the Wisconsin Dells. Really fun time. I stayed at the Kalahari resort and it was really awesome. It had a really cool indoor waterpark with a wave pool. I'd highly recommend it (kinda on the expensive side, though).

I'm very tired. should be back to the norm in the next couple of days.

ahh. XUL and Java

2004-07-20T06:37:00.000-07:00

The JRex project has a Java JAR file that can use the Gecko Runtime Environment and LiveConnect to make a connection between Java and unrestricted JavaScript. This is a powerful ability. A person could make fast XUL GUI applications and still code using a Java, which is a lot more powerful than JavaScript (IMO).

A lot of Java's speed hits would be gone when using XUL over Java's JFC toolkit. Not only that, XUL is a lot easier to read and write than Java's GUI toolkit. It's really a powerful feature that I hope to use in MozCreator once I get JRex up and running on my own system.

New Look

2004-07-19T17:54:00.000-07:00

Just posted the new look for my blog. How do you like it?

Find toolbar

2004-07-19T17:01:00.000-07:00

I can't say that I'm a big fan of the Firefox Find toolbar. I don't understand why this couldn't have stayed as the dialog. The toolbar only creates the bloat of yet another thing to waste space on my screen, and it doesn't appear to provide any new functionality that the dialog didn't provide.

What's worse, is the dialog likes to popup randomly when typing text in blogger (I know that it's probably a blogger bug, but this is a strange problem). That highlighting feature could've been implemented just as easily in a dialog as a toolbar (as far as I know).

I gotta find the bug number and see the logic behind this UI design.

BlueJ

2004-07-18T16:09:00.000-07:00

Just downloaded the BlueJ 2.0 Beta 1 that was released. I've never been a big fan of BlueJ, but it has definetely grown on me. That automatic UML diagram creation is quite useful. The only thing that would help is toolbar button icons :).

The goods are at http://www.bluej.org/download/download-beta.html.

Netscape 7.2?

2004-07-10T20:05:00.000-07:00

What is this? I see the Netscape site covered in Netscape 7.2 downloads (http://www.netscapestore.com/products/software/cdguidebook, http://channels-stage.netscape.com/ns/browsers/default.jsp, http://channels-stage.netscape.com/ns/browsers/download.jsp) but yet the download still goes to Netscape 7.1. I guess AOL really is releasing a new version! I might've been in denial about it for awhile, but I guess it's official!

I guess we'll see if this is a good thing or a bad thing.

Visual J#

2004-07-09T07:34:00.000-07:00

I thought I'd try Microsoft Visual J# .NET 2002, which is strangely for free of the Microsoft website. I found it to be pretty impressive. The omissions of javax.swing components was puzzling, but everything else from java is there (including awt). If .NET could run on Linux and MacOS, it would kill Java just because it's far more flexible for the programmer and has one of the best IDEs I've seen, Microsoft Visual Studio.

Speaking of good IDEs, Brendan Eich told me that it might be possible to make a Visual Studio add-on that could make Mozilla apps (like MozCreator does). It would be interesting if that was possible. Obviously this wouldn't replace MozCreator because it isn't cross-platform but would be nice for Windows users.

Now here's what sucks about J#, the 2003 versions is not free!! Can someone tell me why 2002 is free but 2003 is not? That really stinks. oh well.

Mozilla Firefox 0.9.1 released

2004-07-08T14:12:00.000-07:00

This new version of Firefox contains a security fix that allowed a user to execute a application using a web script and the shell protocol.

No, I haven't updated the help documentation yet. It isn't done :). My goal it to get it in by Firefox 1.0 RC 1, so you can expect it then. My rewritten index will be in there too, so look forward to it (not that anyone really reads that help documentation [considering how many bugs I've found in the documentation that haven't been reported]).

If you're interested in seeing the changes from Firefox version to version, you can see a blog post in SeBlog.

On other news, check out Blake Ross's checkin of the new Firefox Find toolbar, which caught me off guard. I'm not sure if I like the idea of a toolbar button, but the Highlight feature is neat. Go see a toolbar picture at The Burning Edge.

No more IE features ;(

2004-07-08T14:05:00.000-07:00

I'm reading a blog from Ali Ebrahim (http://blog.ebrahim.org/archives/2004/07/08/musings_on_theology.php) where he is describing his discussion with the IE development team at some thing that Microsoft is holding to talk to end users.

I like Microsoft's response to the questions:

Q: When will you support CSS2?
A: We read your feedback.

Q: Will you improve ActiveX security?
A: We're making it recognize the computer it's on, and if there are other bugs, they'll be fixed.

So basically, they're saying that IE will continue to have horrible security and never have standards compliance. Maybe if the developers had a logical discussion on the improvement of IE, something productive might've come out of that discussion.

This is a perfect example of why open-source software ends up becoming far superior to proprietary software development. One of the most obvious examples of this is the success of Apache.

One day Microsoft is going to wake up and realize that its market share is diminishing and that it must actually put effort to competing with its competitors if it wishes to continue having its huge market share. It seems from this discussion that Microsoft is not serious about having a discussion about a user and just claims that because one user wants the feature doesn't mean that another wants it, which is an excuse.

stupid glare

2004-07-07T16:44:00.000-07:00

Can someone tell me why I get a headache after looking at the Help documentation for an hour (just looking, not doing anything) but I don't when I play Unreal tournament or something of that sort?

Maybe I should put a glare protector on top of my glare protector to filter out twice the glare, which might result in half the headaches like when I first installed my first glare protector.

On another note, here's my new Firefox extension manager documentation:

Extension Manager - Installs extensions.
Theme Manager - Installs themes.

What more do you need? If you can't figure them out yourself, then I'm surprised you're intelligent enough to use Firefox.

Hopefully Nilson is around to help write some documentation, because I'm horrible at it :). It's ironic that I'm owner at help systems, yet can't write. Maybe It's my obsession with formatting that helped me :).

MozCreator now on MozDev!

2004-06-28T19:50:00.001-07:00

The MozCreator project has just been set up on MozDev.org! It's located at mozcreator.mozdev.org.

Currently, there is no binaries to install, but source code is posted (with the exception of toolbar images). If you are interested in contributing, email me.

I'm going on vacation tomorrow, but I should have more up when I get back.

MacroToolkit 1.0 released

2004-06-28T19:50:00.000-07:00

Yup, 1.0 is released. Go get it here -> http://macrotracker.mozdev.org/developers_toolkit.php

The size of the toolkit literally doubled, and now includes a cross-platform system registry, data structures, and a new event manager.

Uses Java 1.5 Beta 2, nothing older.

MozCreator source code posted

2004-06-24T08:36:00.000-07:00

Here's the source code to my MozIDE: http://www.trfenv.com/rlk/mozcreator/MozCreator-source.zip (321 K, Zip file)

Requirements to build:

One of the following:
- JCreator IDE 3.0 or higher
- Apache Ant
Java 2 SDK 1.5 Beta 2 (you CANNOT use J2 SDK 1.4)

This package includes all MozCreator source files and the MacroToolkit JAR file (I'm using a nightly version of the MacroToolkit, so I thought that it'd be a good idea to distribute it), MacroToolkit Javadoc information, JCreator workspace and project file, all source files, and ant build script.

To compile, open the JCreator workspace file and choose Build -> Compile Project. Then Build -> Execute Project. If you're using ant, make sure ant is in your path and run the following:

cd c:\mozcreator

ant build

ant run

Because my hard drive crashed recently, I don't have a working Ant installation to test out the script, so if you have problems, email me.

MozCreator Progress 20040624

2004-06-24T08:06:00.000-07:00

A lot of progress was made between yesterday and today. This IDE is much more complete. The only major features not working are program options, XUL GUI creation, RDF generation in the GUI, and gecko embedding.

New features:

Hot Keys added
Fixed exceptions constantly being thrown. It was a bug in Java that caused an exception to inappropriately be thrown when using reflection and private classes.
Automatically generating contents.rdf files are added. Not accessible through the GUI yet.
Added a Browser to the Tab Editor so we can view our code while editing it (the browser doesn't work because Gecko isn't embedded yet).
Status bar component now works. They display the maximum amount of characters and lines (not which one you're on).
Added pretty images to right-click menus.
The Properties sidebar at the bottom now automatically fills in the "File Name" and "File Path" field and tooltips are created to display their contents if they don't fit in the box.
Toolbar buttons now have tooltips.
Projects can be saved in directories that do not exist now (the new directories are created).
HUGE increase in file opening and saving speed. There are also other general speed tweaks.

New bugs:

None!!!!! (at least none found ;))

I will be posting the source code soon, so keep your eyes open for that.

MozCreator Progress 20040623

2004-06-23T14:50:00.000-07:00

I decided to do something similar to what Daniel Glazman does with Nvu and post the progress of my MozCreator creation. MozCreator is a mozilla developer IDE based on Java. It's basically a response to Brendan Eich's Developer Day slides, except that I decided not to do Eclipse (the main reason to use Eclipse appears to be that it is fast and embed Gecko, but it turns out that it's harder to embed gecko in Java in eclipse than through other methods, so I'm not using eclipse :)).

A lot of progress has happened. Here is the new stuff for people who I've told about the project:

The Sidebar tree is fully functional.
Workspaces and Projects are now fully functional, but there is a bug where you have to create the folder the project is saved to in order for the project to be created.

New bugs:

Every time you type a letter a exception is thrown :). This is because the Tab Editor's events are throwing IllegalAccessException's whenever a event is ran for some reason. Hopefully I'll find a lead on that soon.
The workspace file is shown in the project file name.

UPDATE: Brendan told me in an email that he isn't saying we HAVE to use Eclipse, that if there is a better method, to use it. Just to clear up some confusion my comments may have caused.

Eclipse platform

2004-06-22T19:55:00.000-07:00

I recently decided to start development on a Mozilla platform IDE. I was talking to Brendan, and he mentioned that he wanted it to be built on Eclipse so that we can embed Gecko into the app, since they supposedly provided a Gecko control.

Here are my problems with this platform:

The layout managers suck really bad.
The idea that every GUI elements requires a Composite element taken into the constructor is a pain. Java uses a much better method.
The platform uses IE, not Gecko on windows.

I'm definetely considering ditching mozilla.org and just developing this on my own time.

I realize that Eclipse has some great advantages, like running Java apps like a native app, resulting in huge performance increases. But I don't think it's worth putting up with their horrible layout managers. I'm willing to sacrifice the performance. If you don't like it, don't use it ;).

Review: Mandrake Linux 10

2004-06-19T19:50:00.000-07:00

My PC recently busted and cannot boot windows, so I decided to download a copy of Mandrake Linux 10 off of one of the TRF computers and install it on my now-broken PC (even though the windows installer doesn't work, the linux one worked fine). I can say that this version seems much more user-friendly than my old 9 version.

The installer itself was very user-friendly and easy to use. Even though I only downloaded the first CD and had a faulty 2nd CD, the OS was still able to install correctly and run fine. Installing software was a hasle though, because I was somehow using older versions of the libraries (despite that it's the latest Mandrake release) and I couldn't install the new libraries because they depended on libraries on my faulty 2nd CD :). The version came with my favorite apps: Mozlla 1.6, LBreakout, and GEdit. Sadly, XMMS didn't seem to come with the normal distribution, and I couldn't get it to install with the library problem.

During my usage of this OS, I installed Java 1.5 beta 2 and NetBeans. There didn't appear to be any way to make changes to the PATH variable permanent, so I ended up making a text file with the var that I could copy and paste into Konsole (the default KDE console program). NetBeans couldn't detect Java 1.5 until it was in my path, and had difficulties compiling projects. Javac and Java seemed to work differenlty and more randomly. For some reason, running java seemed to cause Konsole to close. I still haven't found a work-around to this problem :). RealPlayer was available to download but was difficult to install the plugin, and once the plugin was installed it rarely worked correctly. Ant was the only program that was extremely simple to install and run.

I ran the new version of Apache web server on this installer and noticed that it installed Apache into an area on the linux hard drive that I cannot modify unless I call manual commands being the root user, so I had to use cp to copy all of my web documents to the server. Once the docs were up, it seemed unusually difficult to get perl, ssi, and php to install. I still haven't found a way to install ASP, still good overall.

Mandrake Linux seems to like to have low volumes for the sound. I have the volume at 100% and it's at about half the volume of my old WinXP install. My video card had some problems with the OS (i'm using a Radeon 9200 w/ 128MB on-board). If I increased the resoluton to 1280x1024, I would get a refresh rate problem with my monitor, and 1024x768 was blurry. Luckly I was able to get some custom resolution in between there that seemed to output fairly clearly.

I tried out Konqueror, the KDE web browser built on top of KHTML. It's a fairly good browser, not quite as good as Mozlla, though. It's basically IE with tabbed-browsing. It still runs unusually slow for animations (try nwtha.com) and doesn't work with gmail. I was hoping that Apple's involvement in KHTML would help the rendering engine work better for animations, but sadly it has not.

The main flaw in Linux still remains: we need to create a standard library across all installations and stop this problem of having about 30+ libraries to install. Even using RPMs were problems because of libaries. In most cases I found it easier just to compile the app from scratch. It shouldn't have to be that way.

Also, linux needs to distribute wine with Linux. Wine is an awesome app that lets you run all my favorite apps on linux, like JCreator, Visual Basic 6, and any VB apps.

Over then next couple of days I'm going to try and get my copy of Unreal Tournament 2004 to install on Linux and if that works, then I'm probably going to stay with Linux :).

IE starts up again.

2004-06-19T19:35:00.000-07:00

Dave Massy, who manages IE feedback and gives it to team members (or something of that sort) has said IE dev has started once again and he is looking for some suggestions.

Well David, may I start with the following:

CSS1, CSS2, and CSS3 support
SVG and MathML support
Get that filter crap out of IE and replace it with CSS3 stuff (like opacity)
Support the WHAT's Web Applications Markup Language

In terms of features, IE needs lots to be better than Firefox :):

Theme support
Tabbed Browsing
A better explorer bar

I personally used to use IE and loved it. But one day I decided to try Netscape 4 and found out how much better it was (and I was using IE 4). That's how I managed to find out that Mozilla existed. If Firefox keeps with its plan of innovation, nothing will stop it.

I would personally like Firefox to support document.all. Call me crazy? fine, but in order for people to flock to Firefox, we need it to be able to render websites correcly. Maybe the WHAT wg need to make a spec for this (to make it somewhat standards compliant :)).

Gmail account giveaway!

2004-06-15T15:51:00.000-07:00

Interested in a gmail account? If you fix 5 or more Mozilla Help Documentation bugs (from Firefox, Seamonkey, or Thunderbird), you will win yourself a gmail invitation! I currently have 5 invitations to giveaway, so if you're interested, go fix some bugs!

More information on contributing at http://www.mozilla.org/projects/help-viewer/bugs.php

Once you have the 5 bugs fixed, send email to rjkeller@gmail.com and I'll send the invitation!

Mozilla Standards

2004-06-13T14:20:00.000-07:00

In order for Mozilla to succeed, I believe that it needs to obtain a better standards body. The W3C doesn't seem to give standards to the advantage of the web developers and obviously isn't getting any respect from Microsoft (I believe that this is the main reason for the WHAT).

The W3C is definitely not the only standards body. ECMA has standardized the JavaScript standard for years now, but we need a web standards body that can quickly accept new ideas and integrate them into the specifications. With the W3C's plan, it takes far too long for web standards to be adopted. New technologies end up rotting away and enhancing web development is only a dream.

For Mozilla's success, I believe that we need to enable support for document.all, marquee, and other non-standard extensions. I know that people say that they are not "standards compliant", but I think the main reason Microsoft supported these technologies back in the 90's was that Netscape was gaining too much market share and Microsoft had to find a way to add new web development features to its browser. Doing this through the W3C took too much time and that was something Microsoft didn't have. If it were to gain market share quickly, it had to think of an alternative technology to make the browser look appealing.

If Mozilla is to succeed as a browser, we need to support these technologies and create our own enhancements. I feel that the WHAT group works more in Mozilla's advantage and to the web developers advantage much more than the W3C does. Mozilla already supports a lot of non-standard technologies (XUL/XBL, marquee) and it would be easy and quick to implement the ideas created by other web browsers to make cross-browser compatibility more realistic. Obviously Microsoft isn't willing to support the standards route. That way was too slow and didn't give their browser the edge they needed.

I'm not saying that standards are bad. They are great and definitely increase productivity and prevent having incompatibilities across products, but the W3C seems to be holding Mozilla back. Most modern web development technologies are beyond the time and scope of the W3C. We need to learn from Microsoft's failures (ActiveX, to name one) and create something superior. XUL/XBL is definitely a huge step down that road. Don't think that I am saying that the W3C doesn't output any great technologies. MathML and SVG are excellent modern technologies, but these things take so long to adopt that by the time they become standards to implement, newer and better ways of doing things have already come out (CSS3 is a good example of this).

My point is, when I see people who (in Bugzilla) say and demonstrate how much they don't support marquee, document.all, or any other proprietary web technology, we need to remember the definition of the "web standard." If the current web standards body isn't benefiting the web developer or the browser producer, then we need to move on to a more efficient organization. The WHAT WG is the solution to this, I believe. It creates standards that demonstrate the technologies of today and not the technologies of yesterday.

Please remember that I am offering this point of view from being myself, a web developer, and what I think will benefit Mozilla the most.

Seamonkey Help File Splitup

2004-06-12T08:12:00.000-07:00

A lot of people have complained about the large help files (mail_help.xhtml is one to name, being 180+K) and want them to be split (bug 246405), so here is my proposal for splitting up the files:

Master Mozilla Content Pack

Mail & Newsgroups Content Pack

mail_help.xhtml
mail_news_settings.xhtml
news_help.xhtml
address_book_help.xhtml

Composer Content Pack

Composer help file with no changes.

Privacy content pack

using_cookie_manager.xhtml
using_password_manager.xhtml
using_form_manager.xhtml
encrypting_information.xhtml
manging_images.xhtml
controlling_popups.xhtml
privacy_on_the_internet.xhtml
preferences.xhtml

General content pack

tools_and_development.xhtml
(all keyboard shortcut files)
managing_profiles.xhtml
using_certificates.xhtml
cert_settings.xhtml

Help on Help content pack

glossary.xhtml
using_help_window.xhtml

And then maybe creating a new directory structure might help also, so here's my thinking on that:

locale/navigator

locale/mail_news

locale/address_book

locale/composer

locale/privacy

locale/general

locale/helpOnHelp

Tell me what you think.

MacroTracker Toolkit

2004-06-11T06:12:00.000-07:00

I posted the MacroTracker toolkit on MozDev. The MacroTracker Toolkit is a set of Java classes I made to help fix some of the short-comings of the JFC API.

gmail accounts

2004-06-09T16:16:00.000-07:00

Responding to Asa Dotzler's blog entry about gmail accounts, I think the easiest way to mass distribute gmail accounts is to advertise them in Unreal Tournament!! I was playing UT2004 and in one of the maps it says you can call 1-800-1Unreal to advertise in the map! I know that it probably isn't a real advertising mechanism, but if it was, it would be an excellent mechanism to get the geekest people to beta test :).

UPDATE: Apparently I missed the point of the whole blog entry, so I now say let's see who can beg the most :). oh well, guess I won't be getting a gmail account ;(.

UPDATE 2: I ended up getting an invite for a gmail account that Tim Meader was nice enough to give to me!! Thanks Tim! So give me a buzz sometime (rjkeller@gmail.com).

help viewer toolbar pic

2004-06-09T15:38:00.000-07:00

Firefox Help on MozDev

2004-06-08T19:49:00.000-07:00

For those of you confused, 90% of Firefox Help development is done on MozDev's CVS servers. The reason is because I'm the only Firefox Help developer with CVS access to the mozilla trunk.

Every once in awhile, I will land the latest development on the Firefox Help trunk into Firefox. If you don't like it, then too bad. So please never ask me this question again, or else I'll just point you to this :).

(rlk's blood pressure goes down)

Firefox Theme issue

2004-06-08T11:24:00.000-07:00

I would like to say that I fully support Ben Goodger's decision to change the Firefox theme. Arvid was sent an email in December about the licensing issue. If after all this time he still refuses to change the license, then we need to move onto something else.

Maybe it's just me, but I don't see what is so illogical about that decision. Yes, the quality isn't quite as good, but we are an open-source project and need to make sure that our technologies can all be distributed under the same license.

Also, you need to remember that the current theme is in a beta state. It's far from completed, so let's wait until the theme is completed and worked on a bit more before we start complaining over its quality.

rlkBlog

2004-06-08T11:14:00.000-07:00

Welcome to rlkBlog! This is my blog (R.J. Keller) where I will post my opinions, views, and proposals related to Mozilla and the Mozilla Help Viewer. I might also post some personal notes from time to time as well :).