Note: The URL selector is stored in the output_buffer - which is currently 520 bytes. Beside all of the URL (apart from a potential "http://") the 'get' and the 'http_preamble' have to fit into that buffer. Therefore URLs mustn't exceed 450 chars. However, we omit a check to avoid further code size increase as most of the time URLs are known to be much shorter anyhow. If the URLs might become large we just leave it up to the user to check their length.
- There was already an explicit code path for something like 'wwww.google.com/' but that code path ended up with the protocol type unset and the port set to 0. Now the protocol defaults to HTTP and the port defaults to 80.
- There was no provision for something like http://www.google.com', rather it was just assumed that the slash after the hostname is always found. Now there's a check if the slash is actually found, and if it isn't found then an empty path is explicitly used.
- Tab chars were partially used with the unusual width of two blanks. I removed them altogether.
- Line endings were inconsistent even iniside individual files. I normalized them.
- I introduced a consistent coding style regarding comment indenting and blank line usage.
- I removed trailing spaces.
- I removed newlines following unnamed labels.
- ...