On 09/15/2014 06:27 PM, Mateusz Kowalczyk wrote:
On 09/15/2014 11:54 PM, Jonathan Paugh wrote:
Hi all,
Using Markdown would be a great idea. And, if Haddock were to support Markdown, and packages were migrated gradually to that, the inconsistency would disappear (eventually).
IIRC, adding Markdown to Hadock was suggested on this list before, and the major argument against it was that Markdown didn't have a standard. Now, it has one, called CommonMark[1]. Barring any (further) copyright issues with the name, that looks to be a great step forward for Markdown.
Regards, Jonathan Paugh
It was more than proposed, it was the original plan of my GSOC in 2013. In the end it went into a different direction because after actually hacking on Haddock I came to the conclusion that Markdown was actually not a good idea. You can find reasons in café archives I'm sure.
Okay, thanks for the correction & info.
There is also the question of who's going to maintain it. Now, that would be a bigger obstacle than any technical ones.
Also, there's the fact that CommonMark hasn't been vetted by the community at large, yet. In case it changes any, this might be a bad time to base a core Haskell component on it. (Aside from these obstacles, I'm all for it!)
We are severely understaffed. Haddock has only two maintainers, Simon Hengel and myself and Simon tends to be quite busy so it's mostly myself, and I also have about a billion projects I want to spend time on. I don't want to maintain another parser, there are more important things to hack. Honestly, it's sad that such a core project is so understaffed. Thanks for all your hard work!
Alas, complaining is not my main aim here. I simply want to point out that after some work, the Haddock parser and the required types now don't depend on GHC or any other libs that don't ship (bar for internal attoparsec dependency) with the compiler. This means that recent Pandoc now has both reader and writer modules for Haddock which means you can go Haddock <=> Markdown if you wish and get some results. I have not tried it myself however. Incidentally, implementing such reader/writer Pandoc modules was my initial submission in 2013.
In summary, if you really want to write Markdown instead of Haddock, use pandoc to convert between the two. I really wonder if the hassle is worth it though.
Regards, Jonathan Paugh