Discussion on:

Choose the right data access strategy when using Crystal Reports with .NET

Do you use Crystal Reports for your applications reporting requirements or do you still generate reports using your own handcrafted code?

Since most of my work revolves around MS Access, I have been using the report writer inside of Access. Once a app is up and running, most of the customer requests are for additional reports or modifying existing reports. Microsoft designed Access so all the forms, queries, reports and programming scripts are part of one massive MDB (or MDE when compiled). Even if I merely move a column 1/4", I have to send the customer a 15 or 20 MB file which needless to say is extremely inefficient. It definitely eliminates sending the report update as a email attachment.

I am going to use a VB.NET program to provide report menus and all user inputs for filtering, sort selections, etc. and have VB.NET run the CR reports. For every CR report there will be two files: the CR report file itself and an INI. The INI file will tell the VB.NET app what the record source is to be, what user inputs are required, what filters to use, what sorts to use, etc. Then the VB.NET app will prompt the user for the required inputs and run the report.

Warner Smidt

We do use Crystal Reports and do use what you call the "push" method (A total misnomer). It works well and makes everything loosely coupled so it is easy to enhance and update as necessary. Contrary to your article, it IS very easy to design the first version since we DO use the underlying data source to create the initial dataset through the magic of VisualStudio.NET drag-and-drop. In fact, after learning how to use the "push" method and a little practice, we've found absolutely no downsides except the very slightly longer set-up is absolutely nothing compared to the trauma involved in making any change to a data source that a Crystal Report is directly attached to (not to mention the flexibility of attaching to different datasources at different times - - like when switching between test and production, etc).

Does a pushed report have to be designed using the strongly typed XSD dataset files or can it be designed using a normal SQL connection and the SetDataSource passed at runtime? If not, what is the best/quickest way to setup the dataset at design time?

Which method should I use, push or pull, when Web Services is involved? Let's say I have published my report as a Web Service. I want to pass a dataset to the report as you would a normal report by using the SetDataSource method, and then setting the ReportSource of the Viewer, etc...My question is, how do you pass the dataset to a report web service from, say, a web page that is consuming the web service? Do you customize the .asmx file to expose some method for passing the dataset in, or do you customize the strongly-typed report.vb object in the web service? Thank You