On Saturday, 26 December 2015 at 20:43:18 UTC, Brian Schott wrote:
> The name lookup code is here: https://github.com/Hackerpilot/dsymbol.

Aye, I'm still just working on figuring out how to use it :)

But since DCD does I should be able to borrow a lil code from there to make it happen.

> If you haven't seen it already, I recommend that you take a look at this: https://github.com/economicmodeling/harbored

Actually, while I have heard of it before, I forgot all about it here. Probably would have been a nicer starting point for this than the dscanner modules I used as guides into dparse.

Oh well. I might steal some from it now! But I still want to do things a little differently. See, I kinda sorta like ddoc and it isn't *fatally* flawed... but it is *fundamentally* flawed and I actually want to get away from it, not implement it again.

I want to have a slightly longer summary. I want to have semantic, properly encoded output. I want pasted links to magically become <a> tags and rich automatic linking.

I do NOT want arbitrary foo: strings at the beginning of a line to start a new section. I do NOT want the current symbol to be highlighted. I do NOT want $1 appearing in the text to do weird things.


So that's going to put a limit on my compatibility with Phobos.

Though maybe, just maybe, I'll use your libddoc on modules that start with std. and core. then do my own things on my modules.....

On Saturday, 26 December 2015 at 21:38:23 UTC, Adam D. Ruppe wrote:
> I do NOT want the current symbol to be highlighted.

Harbored doesn't do that either. The only time I use that feature in ddoc is to disable it.

I've started some parsing of ddoc sections and continue work on the formatter:

http://dpldocs.info/experimental-docs/std.algorithm.searching.findSplit.html

I'm pretty happy with how this is turning out. Going to add more rich info to params later - I realize I can't parse all constraints, but some of them I can - and the ones I can parse cover a *lot* of ground in Phobos!

With any luck, by the end of the day, the params there will say something like:

haystack
    Type: R1, forward range

    The range to search.
needle
    Type: R2, forward range

    What to look for.


Where the "forward range" text will be a link to a conceptual overview of it. Then I will also do automatic See_also: entries that talk about language features it discovered.

This one has an auto return value. I'm not sure my generator can detect this reliably yet but I'm pretty sure it is safe to link to a little doc explaining voldemort ranges too here.


BTW: pred should prolly explicitly document that it requires the function to return bool.

Adam, I wonder could we also have information since what Phobos version particular symbol is available from. For an example, from which version of Phobos we have findSkip() ??

On Tuesday, 29 December 2015 at 14:13:54 UTC, Dejan Lekic wrote:
> Adam, I wonder could we also have information since what Phobos version particular symbol is available from. For an example, from which version of Phobos we have findSkip() ??

Yeah, I was thinking about that too. The info isn't in the doc itself, but could be determined when I update the site here by comparing before and after docs.

So it could automatically say "in the .70 update, I found this function that wasn't there before, so I can add a note `Since phobos 2.70`".


That will take an additional pass and a cross-version database to query during doc builds but it is totally doable.

BTW ddoc also has "Version:" and "History:" sections in the spec we could use to write up changes ourselves...