string = string.split("a").join("b");

I prefer this to using a regular expression. To do something that is intuitively simple, it is not reasonable that the programmer has to introduce the complexity of a regular expression in this case. This implementation of the replace() function in Actionscript 3 is bad because it is not intuitive. The word “replace” in English has the meaning of substituting one thing for another; but it does not include the concept of only doing that substitution in a partial way. So, by having a default implementation that is only partial and not global, the replace() function fails to do what its name implies. Additionally, as a Flex programmer familiar with replace functions in other languages, I may not even aware of the fact that I am not doing a global replace unless I consult the documentation. So the designers of the language have facilitated the introduction of bugs by silently breaking a convention.

//displayView is a Canvas that has been added to the display list
displayView.setStyle(“backgroundColor“, “#FFFFFF“);
	
// here we draw a black line on the Canvas
// but it won’t appear :(
displayView.graphics.setLineStyle(2, 0x000000);
displayView.graphics.moveTo(0,0);
displayView.graphics.lineTo(100,100);

To see your drawing, you need to turn off the background style. As the Americans would say: go figure.

.bubble
{
    background-image: Embed(source=“../swf/bubble_skins.swf“, symbol=“bubble“);
    background-alpha: 1.0;
    background-size: “100%“;
}

The following code will NOT work:

.bubble
{
    background-skin: Embed(source=“../swf/bubble_skins.swf“, symbol=“bubble“);
    background-alpha: 1.0;
    background-size: “100%“;
}

var theta:Number = (Math.PI*2)*(i/NUMBER_OF_SEGMENTS) - Math.PI/2;
var a:Number = 100;
var b:Number = 80;
var x:Number = a* Math.sin(theta);
var y:Number = b*Math.cos(theta);
var point:Point = new Point(x, y);

// container is of type Container
// child is of type DisplayObject
container.addChild(child);
container.addChild(child);

you get the error message “Error #2006: The supplied index is out of bounds.” That’s not a particularly helpful message, because it’s a very general error that is generated inside the Container.addChildAt() method; even though you are not explicitly defining a child index, the error is related to indexing of the array that holds children of the container. You would get the same error if you tried to use container.addChildAt() and specified an invalid index.

A better error message for this case would be something like “Duplicate child object not allowed in single instance of Container”.

.content{font-family: arial;}

but this will work:

.content{font-family: Arial;}

Unfortunately, none of the native Flex components have layout algorithms that are suitable. They mostly implement absolute positioning or a simple vertical or horizontal “flow” layout which can fit elements into the parent container, but without any resizing.

Recently I have found a really nice implementation of layout within Flex called the Adobe Flex 3 Dashboard sample. It implements a “pod layout” which handles all my requirements EXCEPT for the fact that the individual elements (”pods”) cannot be given an initial/default width and height - they simply scale to each take up an equal amount of space within the parent container. The drag and drop functionality also has some nice animated effects.

What is worse is that no error messages or warnings are shown, so you have no idea that anything has gone wrong. The only way to detect the problem is to manually compare the imported data row by row with the import data source spreadsheet.

This problem persists no matter what data type you might choose for the destination column.

Searching around in the forums for this issue, I have come across some talk of changing the connection string used to the OLEDB connection to include “IMEX=1″. That works fine if you are building a SSIS Integration Services package in the full version of SQL Server 2005. But there is no way to do anything like that in the SSIS Import Wizard - no access to connection strings and no way to choosing different providers for importing Excel files.

So I have found my own workaround which is quite simple. What I do now before importing any Excel sheet is to check the first few rows. If there are columns that have empty values in the first few rows, I check further down the sheet. If the same column has data further down, I know that I am in danger of losing it during the import. So I create a dummy first row of data that has all columns filled out with representative data. For example, if a column contains integers, I might put a 1 in the dummy row; if it is string data, just a few characters. This seems to make the import process less cavalier about ignoring data, and I find that all columns now import correctly for all rows.

After the import, I run a query to delete the dummy row of data.

The workaround is fairly simple, but I must say this is a horrendous bug. Not only is it a high impact bug but the user has no feedback that anything has gone wrong. This is on top of the problems I have had with arcane error messages when an import fails because of bad data; and problems with row delimiters when importing CSV files. I have lost confidence in the SQL Server SSIS import now. The old SQL Server 2000 DTS seemed much more stable and reliable. SQL Server 2005 has been generally good to use, but not having a reliable way to import data is a severe handicap and makes me less likely to continue using it.

By the way, you can find an excellent tutorial on how to use the new SSIS Import/Export Wizard in SQL Server 2005 here.