|Posted by Walter Bright|
in reply to Derek Parnell
Posted in reply to Derek Parnell
"Derek Parnell" <email@example.com> wrote in message news:firstname.lastname@example.org...
> On Thu, 29 Sep 2005 21:48:27 -0700, Walter Bright wrote:
> > "Derek Parnell" <email@example.com> wrote in message news:firstname.lastname@example.org...
> >> 'bool' comes out as 'bit, 'size_t' comes out as 'uint', etc...
> >> If I wanted 'bit' and 'uint' that's what I would have used in the code.
> > I know. It's on the list to fix.
> I'm sorry if I'm sounding a little gruff; its not personal.
> I think you've done yourself, and all D users, a great service by developing DDoc. Sure, I have a couple ideas for enhancing it but I'll
> fire until it stabilizes and gets some real-world usage.
Hey, no prob. The reason for the alias problem is the semantic pass has already been run, and aliases have been replaced by their real types. Running Ddoc before the semantic pass means that things like class heirarchies will be unknown. So a copy will have to be saved of the original type.
Right now I'm working on adding color syntax highlighting for code sections. It's actually easier than I thought. I just hook the highlighter into the lexer, and voila! It gives me accurate start/stop for each token type. I've never seen a C++ syntax colorer that gets it 100% right, because it requires a C++ compiler and compliant preprocessor to do all the edge cases correctly. But Ddoc will do it 100%, by leveraging the actual D lexer.
The irony in all this is that if I wasn't so lazy, I'd never write Ddoc. I'm too lazy to do the meticulous and boring work of tweaking all the docs by hand, and let's face it, HTML is just not pretty, so I write a program to do the dirty work for me. And it is kinda fun watching it turn routine code comments into eye candy <g>.