School: Rensellear Polytechnic Institute

Major: Computer Science

Year: Senior

E-mail: foutzl@rpi.edu

Webpage: http://www.rpi.edu/~foutzl filled with worthless comic book trivia and some stuff I really should take down.

Goals: World Domination or tenure, whichever comes first.

My mentor for the summer is:

Name: Sandy Mamrak

School: Ohio State University

Department: Computer Science

Area of Research: Database

Webpage: http://acuity.cis.ohio-state.edu

E-mail: mamrak@cis.ohio-state.edu

Research Project

My project for the summer is to compare two input form generators, the Automatic Form Generator(AFG) created at OSU, and Architemp, created by Dulcian Inc. Input forms for databases are just GUI interfaces for adding data to the database. The AFG is a Java based program that takes an XML spec created by the developer and creates a form program from the spec. Architemp provides the user with a generic template which she can edit using Oracle Forms Developer to create the form she wants. My duty is to compare the two and figure out which one is more efficient at creating useful input forms.

Weekly Journal

Week 1

This week I worked on several practice assignments meant to help me become familar with the database and to give me a refresher on Java, SQL, Unix, and PL/SQL. A list of these assignments is located at http://acuity.cis.ohio-state.edu:8888/NewStudents/index.html.

Things I learned this week:

• 50 percent of all Unix errors are due to permission problems (the command "ls -l filename" will show you permissions on a file or directory).
• The other 50 percent are classpath errors (use "echo $CLASSPATH" to see the classpath, "setenv newpath:$CLASSPATH" to add a new path to the classpath, and "fine / -name filename -print" when all else fails to search the entire system for your file).
• Java programs run better when compiled.
• While books never tire of describing every tiny detail and obscure command in a language, they usually forget to describe the small but very important things, like that PL/SQL programs must end with a "\/" or they will not run.

Finished up on the practice assignments and began reading up on Oracle forms in preparation to download and install Oracle Forms Developer and Dublin Achitemp.

Things I learned this week:

• Two cups of coffee will not keep you awake while reading the install instructions on Oracle Developer, but one Red Bull will.
• It is good to be root (the all powerful user identity on Unix and other systems, also the name of the top of the file system tree).
• Remember to remove all humorous error checking output before distributing a program (I seriously doubt elves ate the file I was trying to get).
• Documentation suffers the same aversion to small but important things that books do.

Began to download Oracle Developer, but when it was discovered that I could ruin the database if I entered bad data for the set up variable it was decided to call in a more experienced person to do it. So I was given the task to update some download scripts to the database.

Things I learned this week:

• Perl is one the of best languages for parsing text and data.
• Most of industrial programming is taking something someone already wrote and updating it.
• Because most programming is updating existing programs, liberal comments of a program's code is a must.
• "desc table or view name" is the best command in SQLplus ("desc" gives the types and names of columns in a table).

Began to download the tared file of Oracle Developer onto the server sagacity because it has plenty of empty memory. Turns out the Oracle Developer install file is very big, bigger than the memory available on sagacity, so next week it will be downloaded onto the computer science system and installed later on sagasity. Till then I am to download lots of information from the Mayo Clinic into the database.

Things I learned this week:

• It is good to have four versions of one database, three to mess up, and one to keep running (hopefully).
• While it is heartbreaking to finally get a program working and then have to try your best to break it in error checking, it has to be done.
• 1000 entries from the Mayo Clinic is a great way to test a download script.
• When asking a user for input it is best to do as much instant error checking as possible, never underestimate the users ability to screw up royally.

After downloading the install file for Oracle Developer onto the computer science system we began to install it on sagacity. To ensure that there would be enough memory we deleted an old obsolete version of Oracle. Turns out there still was not enough room to install Oracle Developer Forms, so an order was placed for an eight gigabyte hard drive for sagacity.

Turns out that old obsolete version of Oracle was actually important. Despite everyone's best efforts I still managed to destroy the database. Till then I was asked to create three html input forms and cgi scripts to prevent the error that happened with the Mayo Clinic data from happening again.

Things I learned this week:

• It is very good to have four versions of one database, the deletion of the old version of Oracle destroyed the three test databases, but left the one for the customers up and running.
• Never underestimate the users ability to mess up, and error check with that in mind.
• When doing error checking try to think of common mistakes, like formating dates wrong.
• Write error messages even for things that should not happen, like missing files or command line arguments, because the new program just written will most likely be changed by someone else in the future.

Still waiting on the new disk for sagacity. Till then it is my job to write cgi scripts to go with the input forms. Unfortunately the system does not have the CGI.pm library of common cgi built in functions. So I go to download it and it turns out our version of Perl is too old to support it. So I go to download a newer version of Perl but it turns out that our C compiler is not working so Perl cannot be compiled. So I go to download C, realize that is just too much work, and learn to write a cgi script without CGI.

Things I learned this week:

• A list of parameters sent from an input html form can be acquired with the command "$ENV{'QUERY_STRING'}".
• When misspelling variables in Perl try to be consistent in your misspelling, or you will get the strangest and most difficult to track down run time errors.

This week I was asked to download and compare some XML editors. This was good practise for when I compared the AFG and Architemp. I searched through several sharware and demo engines and compared everything from Merlot to XML Spy. I came to the two conclusions that Merlot was the best all around xml editor and that there is a lot of useless software out there on the web.

Things I learned this week:

• People really will pay $600 for an XML editor.
• A lot of people have a lot of spare time, and will use this spare time to make some really nice free software.
• Microsoft truly is trying to take over the world (I tried to print out a table I created using the Internet Explorer html editor from Netscape and the printer encountered an error, when I tried printing it from Internet Explorer there was no problem).
• Platform independence is a wonderful thing, and by extent Java is a wonderful thing.

This week I had to write a query for all the information that was put into the database using my input forms. This would not have been such a bad job except that the query was insanely complicated. I kept having to rewrite the query because the one I just created was so complicated that it took ten minutes to execute. I finally discovered a few new SQL commands and was able to write a working and efficient query.

Things I learned this week:

• The SQL decode function is a lot like an if statement.
• When joining tables make sure you put some constraints on the joining, or you will end up with 23416 rows returned from two tables with only a hundred rows each.
• Joining 10 tables takes a lot of time.

Yay, the sagacity disks finally came and Oracle Forms Developer was installed. I finally got to begin the project I was supposed to do at the beginning of the ten weeks. My job is to create an input form for the Molecular Pathology table and compare the creation process with when I created an input form for the same table using the AFG. Does not sound that hard but Oracle Forms is very complicated, and all the bugs in the Architemp demo did not speed up the process.

Things I learned this week:

• Error filled demos do not impress prospective customers.
• Neither does documentation written for an earlier version of the demo.
• While flexibility is nice, 50 different variables for a single button is just to complicated.

This week I write up a report on my comparison of the AFG and Architemp (AFG won), fix a small error with the input forms, and do other things in preparation for my leaving. All in all, this has been a great job.

Things I learned this week:

• Chmod 775 is my friend (gives people in your group write permission, so they can delete old files after you leave).
• So is chown (changes the owner of a file).
• In the end, simplicity is better than flexibility (which is why the AFG won over Architemp).
• I really do want to do this research stuff for a living.

Due to the memory problems I did not do one major project this summer, but rather lots of small projects that took about a week each. These projects did all have a common theme, they all involved inputing data into the database. I had the oppertunity to work on the data input process from creating a GUI for the user to input data with, to formating the raw data, to finally placing it into the database.

My list of projects is as follows. I created three html forms for entering data and wrote the cgi scripts to handle the information. The cgi scripts only place the information into text files, so I updated several existing programs so they could download the raw data into the database. So the users could view the information they had inputed I wrote a query that pulled up their information.

The database is an ongoing project that is constantly being improved, rewritten, and expanded. While I was comparing the AFG with Architemp others where working on creating a GUI for creating the XML specs nessasary for the AFG to create a new input form. This GUI will make the AFG even more user friendly and will allow it to surpass Architemp in the one area where it previously lagged behind. New tables and fields where also being discussed for the database, so the input forms, perl scripts, and queries that I wrote will have to be updated to reflect these changes. Hopefully I commented these programs well enough so that the next person to update them will find them easy to understand.