There are two parts to solving this problem. When the Data Lump is initially read we need to accumulate and tag comments for later use. Comments can be tagged to a data structure or they can be tagged to a field within a data structure. When parsing the data lump the SDK will need to track whether the last element added was a data field within a structure or not. If a comment is encountered then it can either be tagged back to the last data field or, if not currently parsing a structure then comments can be accumulated and tied to the next structure that is encountered.
I'm kind of old school about my coding practices and I don't like to use a lot of extra memory when it isn't needed. In the case of tracking comment tags the lightweight solution is to not have every element carry a tag to a comment field but to have all of the comments that are encountered carry a memory pointer back to the element they belong to. When the API is writing the data lump back out to disk then when each element is written - where an element is defined as either a structure or a field within a structure - we will see if that elements memory address matches the memory tag on a comment. If so then write the comment field out as appropriate. Appropriate for a structure is before the structure begins and for a field it means a tab after the data record is written followed by the comment. All comments are written C++ style with a two forward slash, //, preface. Comments that exceed a reasonable line length will be continued after a newline is written with record comments that exceed a single line being prefaced by multiple tabs on the newline.
Part 5, the last section, entitled writing the Data Lump API, is coming soon,
L
No comments have been provided.
Your Name: | |
Your Location: | |
Country (flag): | |
Article Rating: | |
Your Comments: | |
Security check * |  |
style: inset; border-color: black black black black;
border-width: 3px;
background-color: #000000">
