Read the conversation — then try your phone

Take the URL apart, curl -v your server, map a status code — then open it on the phone.

You will be able to

Split http://192.168.1.42:8000/ into scheme, host, port, and path.
Run curl -v http://localhost:8000 and spot request lines (>) versus response lines (<).
Map 200, 404, and 500 to who has the problem — then open http://localhost:8000 on your phone.

The server is still running. On the laptop, open a browser.

Type http://localhost:8000 and press Enter.

Your ~/projects/first-site/index.html should load. That string in the address bar is a URL.

A URL is the address of a unique resource.

You are looking at your own page, served by the process you started. Next you will read the conversation that produced it. Then you will type the same address on your phone.

Take the address apart

Read http://localhost:8000/ from the left.

The first piece is the scheme. The scheme is the protocol the browser must use. Here it is http.

:// only joins the scheme to what follows. It is not a name.

Next comes the host. The host is which machine is being asked. Here it is localhost.

Then a colon and a port. A port is the gate on that machine. Here the gate is 8000.

You have to type :8000. Browsers hide the usual HTTP gate, which is port 80. They also hide 443 for https. 8000 is neither, so the number stays in the address.

The last / is the path. The path names the resource on the server. A lone / is the default page — your index.html.

AnatomyThe address on your laptop

Click any piece of http://localhost:8000/ to name it.

://:

schemehttp

The protocol the browser must use.

hostlocalhost

Which machine is being asked. An IP address may replace a name here.

port8000

The gate. You type it because 8000 is not port 80.

path/

Which resource. A lone / is the default page — your index.html.

Four named pieces. The punctuation only joins them.

Cover the address with your hand. Uncover one piece at a time and name it. Scheme. Host. Port. Path.

The host is usually a name. An IP address may replace that name.

Split http://192.168.1.42:8000/ the same way, piece by piece.

scheme: http
host: 192.168.1.42
port: 8000
path: /

Same four pieces. The host is written as numbers this time.

A piece after # is a fragment. Add a # and some letters on the end if you want to see one. The fragment never reaches the server. The browser keeps it after the page arrives.

Print the conversation

HTTP is a protocol for fetching resources such as HTML documents. It is a client-server protocol. The client starts every exchange.

Your browser is a client. curl is a client too. The process on port 8000 is the server.

Clients and servers swap discrete messages, not a stream. The client sends a request. The server answers with a response.

HTTP is meant to be human-readable. You can print the words and follow them.

Every message has the same four parts.

A start-line.
Optional headers — one per line.
A blank line, which ends the headers.
An optional body.

A request’s start-line is the request-line: method, then the path, then the protocol.

A response’s start-line is the status-line: protocol, then a three-digit status, then a short phrase.

The server is occupying one terminal. Open a second one.

Against your server — the one serving ~/projects/first-site — run:

curl -v http://localhost:8000

-v means verbose. curl prints the conversation around the page.

Scan the output for two marks.

Lines that start with > are the request. You sent those.

Lines that start with < are the response. The server sent those.

Ignore anything else for now. Find the first < line. That is the status-line.

On this server it reads HTTP/1.0 200 OK.

The body that follows is index.html. Same file you opened as file:// in Unit 1. This time it arrived as an HTTP response.

SequenceOne request, one response

curl (client)your server
a request for /curl -v prints this block with > — you sent it.
your servercurl (client)
HTTP/1.0 200 OKThe status-line. curl -v prints it with < — the server sent it.

Two discrete messages. > is the request. < is the response.

Below the status-line sit the headers.

Content-Type names the media type of the body. Yours should show Content-type: text/html. That is the HTML of index.html.

Content-Length names how many bytes follow. It is a count, not a grade.

Then a blank line. Then the body — the same markup the browser painted.

Scroll back up the verbose output. Find a > line that names a method and a path. That is the request-line. Find the < status-line again. One went out. One came back.

If nothing is listening, there is no status-line. curl never starts an HTTP conversation. It prints curl: (7) Failed to connect… Could not connect to server and stops. That is a closed gate. It is not a 404.

Three numbers, two owners

The status is a three-digit code. The first digit is the class.

First digit	Class	Who has the problem
`2` (`2xx`)	success	nobody — it worked
`4` (`4xx`)	client error	you, the client
`5` (`5xx`)	server error	the server

200 means the request succeeded. That is the page already in the laptop browser.

Now ask for a path that is not a file in ~/projects/first-site. Keep -v so you can see the marks. Read the first < line.

404 means you asked for a missing path. The server answered. It is working. It does not have that resource.

500 means the server broke. An unexpected condition stopped it. The path may be fine. The machine on the other end failed.

A 404 is your miss. A 500 is the server’s miss.

Reload http://localhost:8000 in the laptop browser. That request is still 200. A 404 on a missing path does not take the real page down.

Type it on the phone

Keep the server running. Do not change the URL on the laptop.

Pick up your phone. Open its browser.

Type http://localhost:8000 exactly. Go.

Wait until it fails. Do not edit the address. Do not switch apps to “fix” it.

It fails.

Wait — the laptop shows the page, and the phone cannot open the same address?

That feels like a typing error, or a broken phone. It is neither. You copied the working address. localhost means the machine that is asking. Unit 4 will get this file onto the phone.

Do not change the URL yet. Do not stop the server.

You have read the HTTP conversation for this file. The phone cannot load http://localhost:8000. That is the wall.

Flashcards

In a URL, what is the scheme?

The protocol the browser must use. In http://localhost:8000/ it is http.

1 / 5

Quiz

Question 1 of 3

A 500 means you asked for a page that does not exist.

False True

Show answer

Answer: False

The opposite is true. A missing path is 404 — you asked, and the server is working. 500 means the server hit an unexpected condition and could not fulfil the request. (objective 12)

Question 2 of 3

Split http://192.168.1.42:8000/ into scheme, host, port, and path.

Your answer

Show answer

Answer:

scheme http, host 192.168.1.42, port 8000, path /

A full-credit answer shows: A strong answer covers all four pieces: scheme http, host 192.168.1.42 (an IP in the host slot), port 8000, and path /.

:// joins the scheme to the authority. The authority is host plus port, so 192.168.1.42 is the host and 8000 is the gate. http is the scheme, not the machine. The final / is the path, not decoration. An IP may replace a domain in the host slot. (objective 10)

Question 3 of 3

You run curl -v http://localhost:8000 against your server. Which lines did the server send?

Lines that start with > Lines that start with < curl: (7) Failed to connect… Could not connect to server Only the HTML body, with no prefix

Show answer

Answer: Lines that start with <

curl marks the response with < — status-line, headers, and the start of the body. > is the request you sent, so those lines are yours. curl: (7) is what you get when nothing is listening: there is no HTTP conversation at all. The HTML is part of the response, but it is not the only thing the server sent, and verbose mode prefixes the head with <. (objective 11)