Web tools for mod_ruby - Dave Thomas

ruby-talk
1

Hello all,

This is mostly a message for Dave Thomas. I just read your presentation
"What I did for Summer Vacation" (can’t remember the exact title). I am
curious, you talk about how you ran into a few problems with sessioning. I
am wondering if the classes you developed can be posted in the RAA, or if
similar classes are now in existence.

I have been a PHP coder for a long time now and while its OO environment not
nearly as pristine as Ruby’s, the convenient functionality that it makes
available to web programmers and its high performance is tipping the scale
in its favor for web development. But I’m really interested in branching
out and trying web development on Ruby. However, these types of tools make
an enormous difference. Right now, mod_ruby appears to be very much in its
infancy with little activity on the modruby.net web site.

Thanks for your time.

Carl Youngblood

2

Carl Youngblood wrote:

I have been a PHP coder for a long time now and while its OO environment not
nearly as pristine as Ruby’s, the convenient functionality that it makes
available to web programmers and its high performance is tipping the scale
in its favor for web development. But I’m really interested in branching
out and trying web development on Ruby. However, these types of tools make
an enormous difference. Right now, mod_ruby appears to be very much in its
infancy with little activity on the modruby.net web site.

One of the constraints imposed by my clients is that I couldn’t use
cookies, so I implemented sessions using URL rewriting. This wasn’t
really anything to do with mod_ruby; it was just some stuff in my code.
Whenever I needed a link or button on a page that was supposed to call
back to my page, I invoked a method

  {
     "ok"  => url(:handle_ok),
     "help" => url(:handle_help, "register", user_details),
  }

This method did two things. It saved the name of the method to be
called, along with any parameters, in a dispatch table. It then
generated a unique number which encoded a session ID and the index into
that dispatch table. This is the URL given to the browser.

Then, just before my code finished handling the original request, it
persisted the current session (which includes the dispatch table) as a
serialized object into a row in a database table.

When the user clicks on one ofthe links, control comes back to my code.
I decode the URL to get the session key, then use it to read in the row
from the table. I deserialize that, then reconstitute all by session
data and the original dispatch table. I then decode the index in that
table out of the URL, and invoke the appropriate method using any
parameters that were perviously saved.

I’m afraid I can’t post the code: it’s currently commercial. However, it
isn’t big: 2-300 lines at most.

Cheers

Dave