Notes for Thursday, July 3 Meeting

• PDF parser worked for test file that he used but the test file turned out to be blank and also not linearized (optimized for fast processing) and so other linearized pdfs caused seg faults.
• Planned to fix this, and, as of the writing of these notes, has done so : )

Steve's Report:

• Will use the Xerces Java XML parser that comes with Java 2. However, it only parses XML data, not schemas, so a schema parser will have to be written as he couldn't find any online.
• Creating schemas by hand for now, one as the metadata attached to a specific form and one as the schema for the user's personal data on his or her computer. He will parse these two together to obtain the mapping from the pdf field names to the user's personal data.
• Wrote a program to merge two schemas. The program checks for differences and merges in new fields.
• The goal is to make every field path in the schema hierarchy unique and, during the merging of two schemas, to take the union of the two trees.
• Kris pointed out that the occassion of two leaves in the schema tree colliding is not a problem because if the paths are equal then the fields have the same semantic meaning.

Contributions from everyone (sources unknown):

• Need a java applet to interact with the user. This applet would produce warnings and requests based on policies that the user has selected regarding time-to-live, sensitive data, etc. It would make it easy to interact with the form filler. Ultimately it will be fully interactive but for now we should present the user with the fields to be filled in automatically and those that the user must supply.
• The GUI should have a list of preferences in the form of a drop down list so the user can customize the interface to his or her liking.
• Neeraj is working on the GUI to implement this interface and he or someone else will eventually design a GUI that will make creating and merging the schemas easy for us.
• The basic process to fill a form is: parse the pdf, get the field names, get the user data, then fill the form.

Anthony:

• CVS will be set up soon so that we can check out files, work on files concurrently, and have access to each other's source code.
• For now check out the CVS homepage (http://www.cvshome.org/) and browse the manual to (re)familiarize yourself with it.
• CVS notes: when checking in a binary file use the -kb flag to signify that the file is binary otherwise it causes problems. When updating use the -d flag to create the relevant directories.
• Our main server is Dwight but is not set up for our use yet.