Using SharpZipLib for this seems pretty complicated. This is so much easier in DotNetZip. In v1.9, the code looks like this:
using (ZipFile zip = new ZipFile())
{
zip.AddEntry("Readme.txt", stringContent1);
zip.AddEntry("readings/Data.csv", stringContent2);
zip.AddEntry("readings/Index.xml", stringContent3);
zip.Save("Archive1.zip");
}
The code above assumes stringContent{1,2,3} contains the data to be stored in the files (or entries) in the zip archive. The first entry is "Readme.txt" and it is stored in the top level "Directory" in the zip archive. The next two entries are stored in the "readings" directory in the zip archive.
The strings are encoded in the default encoding. There is an overload of AddEntry(), not shown here, that allows you to explicitly specify the encoding to use.
If you have the content in a stream or byte array, not a string, there are overloads for AddEntry() that accept those types. There are also overloads that accept a Write delegate, a method of yours that is invoked to write data into the zip. This works for easily saving a DataSet into a zip file, for example.
DotNetZip is free and open source.