Carlos Maltzahn wrote: > > Except the transfer of XML from tool to tool. Yep, we're going to get killed on the terminology in this project. When you say "transfer", you mean reading the XML file from disk and then "streaming" it to the next tool, without writing back to disk? As you wrote below, we shouldn't have to do this if both tools are on the same NFS. But we will need some mechanism for XML transfers between NFS's. > It might make sense to make a distinction between streamed tool input and > input files. Right. So workflow info is streamed and biological data (XML) is read/written from/to a file. > If a tool has to have access to the entire result of the > previous tool before it can do anything useful and the tool runs on a host > that shares the same NFS with the host of the previous tool, it doesn't > make sense to transfer any data: all the tool needs is a pointer to a file > that contains the result of the previous tool. Yes. > On the other hand, if the tool is able to process a stream of data, the > entire process can be pipelined which saves a lot of time. Now by "pipelined", you mean one tool starts getting input before the other tool is even finished with the data...for serialized analyses? I agree we need a glossary. I'll get started on one, but some of the terminology I use may not be "correct" and will need to be changed. Jeff -- J.W. Bizzaro Phone: 617-552-3905 Boston College mailto:bizzaro at bc.edu Department of Chemistry http://www.uml.edu/Dept/Chem/Bizzaro/ --