VS 2005 - Compressing A Dataset Converted To XML Using The GZipStream Class Added In 2.0

I am compressing a dataset converted to XML using the GZipStream class added in 2.0. There are a few things bothering me about this. Here's the code I am using to decompress the XML file:

Using inFile As New System.IO.FileStream(flnm, IO.FileMode.Open, IO.FileAccess.Read, IO.FileShare.Read)
Dim zStrm As New System.IO.Compression.GZipStream(inFile, IO.Compression.CompressionMode.Decompress)
Dim buff(999) As Byte

[CODE]...

This seems inordinately awkward, as I am taking the stream, reading it out into byte buffers, then writing those byte buffers to a memory stream that is then passed to the dataset.ReadXml() method. The reason for this seems to have to do with what is happening in the line colored red. That line is adding a closing > onto the end of the stream. Without that, ReadXml complains about an unclosed dataset. I found out that the > was missing by writing the stream into a string and pasting the string into WordPad. Without that missing >, it appears that I wouldn't need to be using that memory stream at all, and could simply pass in the GZipStream to ReadXML, but as far as I can tell, it's impossible to look at the GZipStream to be sure that the > is not there, and it may be impossible to add it (I haven't tried).

So why is the > missing? Here's the code used to create the file that is being read into the filestream in the above

VB Using ms As New System.IO.FileStream(flnm, IO.FileMode.OpenOrCreate, IO.FileAccess.Write, IO.FileShare.Write)
Dim zStrm As New System.IO.Compression.GZipStream(ms, IO.Compression.CompressionMode.Compress)

ds.WriteXml(zStrm, XmlWriteMode.WriteSchema)
End Using

The real kicker to all of this is that while I was dissecting the first chunk of code, I added a line to check for cnt<1000. My intention was to see the last buffer that was being written into the memory stream to see whether the final > was actually there, or not. To my surprise, I reached the cnt<1000 twice. That means that one read did not read a full 1000 bytes, but read only 675. The next three or four reads were a full 1000 bytes, then the final read was just a few hundred bytes. It is my understanding that Read will return a full buffer if there are that many bytes left to read, and there were, so why did it return only a partial buffer? That suggests that there might be a hole in my XML, though with a half meg file, it's a bit much to read through it looking for something missing. However, the XML that was returned was turned into a dataset that I then merged into some existing data using code that has been working with uncompressed XML files, and I got an error that I should never have received. That may indicate that the XML was, in fact, missing something, though I will have to study the matter further to figure out whether or not that is the case. My major question has to do with the missing '>' at the very end of the file.