HTTPD

As the site abstractions worked well in a test yesterday, I today started to implement a test server for the new HTTP response. The very first compile showed some missing functions in the response and then I could actually make a first test run. The code seems to work and stopped where it was supposed to.

With those things finally out of the way, I continued work on reorganizing the code so I can support responses having their own send buffers. Today I at least added a function so the connection object can query if the response uses its own buffers. If it does it branches into a different send mechanism which is not implemented yet. But at least I am making progress and hope to soon have the first step (uncompressed response) implemented. Then I have to add compression and finally chunked transfer.

I have implemented a first primitive path handler. This one gets a full path and a string to send as response to a request. I have also implemented all the parts necessary to actually find the right handler and send the request. I have even implemented a simple 404 ("page not found") response. A simple test server using these new classes was also created and it even works. So the basic abstractions for a site seem to work well.

The next thing to implement is a path handler for dynamic pages using the new response type. And then I can finally finish the new response type.

I considered creating a new test server for the new response in the way I have done it before. But I think it is time to build the next layer on top of the raw connections. That layer is responsible to schedule the requests to the code which can deliver the right response. Basically it has to look at the path component of the called URL and decide what page to send as an answer.

I have started to build those abstractions, currently I have the abstraction for the whole site (named 'Site') and the start of the base class defining the handlers for path components. I will also derive some simple handlers which for example return static pages, images or scripts. When it is finished I will be able to serve more than a single page from my http-server without effort.

I had been looking into the code last week end to find out, how to support the next evolution in the HTTP responses. I didn't see a good way and so I didn't do anything, I just started to think about how to do it. Today I have started to actually change the code so I can add a new kind of response. Currently the HTTP connection object creates a send buffer and the responses write their contents to that buffer from where it is sent. The new responses I plan to implement have to have the ability to handle their own send buffers.

I think it is possible with my current design and doesn't even distort it. But I needed to make quite a lot of small changes and reorganize the class structure. Most I should have done today and so I should be able to start implementing the new response. I will probably take some time to implement the part gathering the data. Then I will see, if the changes I still need to make to send that data are possible. At least I will have a much better understanding of the problem.