File Repositpory Project

Problem statement:

Milestones:

• October 23-24: Set up web-interface for member to enter conference page, the page should list conference members and information for file uploaded to the conference.
• October 25-26: Create the database table for storing the File repository information in the SQL database.
• October 27-29: Implement the file upload, update, deleting. The file list should support ordering by each field.

SQL Database:

We store all the file information in a SQL database named file_repository. Following fields can be considered in the file repository table.

conference	varchar	The id of the conference
creator	varchar	The id of the owner
file_id	int	The primary of the file, automatically incremented.
file_name	varchar	name of the file
file_type	varchar	text/plain or application/postscript(by Icon)
file_size	int
file_description	varchar	The description of the file information.
file_access?		Public or restricted
date_created	datetime	Create time for the uploaded file.
date_modified	datetime	latest modified time of the uploaded file.
Edit?		Update/Delete, creator(update/delete), moderator(delete)

Henning's comments

Note that you should be able to handle a whole range of filetypes,
including PowerPoint, PDF, plain text, HTML, Word, etc.

Access should be:
- public
- member (any conference member)

Use an enum.

Instead of Edit?: should note whether the file can be overwritten or
deleted by
- moderator
- member (any conference member)

Also enum.

> I don't know whether it has been discussed earlier, but why do we need to
> store the files in the database (why not files in the file system).

Longer term, I'd like to store the file data in the database, as that
avoids access rights being different for the file system than for the
database. Also, the mySQL database allows replication for load
balancing, which wouldn't work if there are files involved. (We would
then have to assume that all replicas have a shared file system.)
Finally, if we want to introduce directories and file versioning, this
requires some hackery on the file system part.

Also, if you store files on the server, you now have to synchronize the
mime.types table to make sure that the file extension produces the
correct content-type.

Finally, it's easier to write a file access logging mechanism that way,
without having to rely on the web server. (This is far down the priority
list, but I'm planning ahead.)

However, currently the Tcl mysql library cannot handle binary data. I
had asked Kundan to take a look at libfbsql; I don't think this is hard
to fix.

By Huwei Zhang, Computer Science Department of Columbia University.