On 10/07/2011 10:24 AM, Robby Findler wrote:
On Fri, Oct 7, 2011 at 11:15 AM, Neil Toronto<neil.toro...@gmail.com> wrote:
Here's the thing. When I first looked into it, I wasn't making a collection;
'plot' was just a directory on my hard drive. I tried using
'scribble/srcloc', and it kept not finding modules. So what I think is true
about it are often assumptions and inferences.
For example, I think that it wouldn't give me the freedom to present the API
in the order/grouping I want it in, for the cases where the best
order/grouping in the docs is different from the best order/grouping in the
code.
I think there are hooks in there to make this pretty flexible. The
framework docs use them to slice up things and move things around. You
might take a look at how it is set up and see if something like that
would work.
I'll take a look. Thanks!
When a syntax object is compiled and written to a .zo file, all that
information is stripped. You have to write use another function at
compile time that walks over the syntax objects and records the source
locations and then saves them in some way that is not a syntax object
literal. Then restore them.
I was afraid of that. Probably not too hard, but not a cakewalk, either.
There may be some library that I don't know about that helps with this.
Robby
I looked around and didn't find anything immediately useful either, so I
wrote a 'serialize-syntax' and 'unserialize-syntax'. Serializing doesn't
store lexical context, so unserializing takes a context argument - it
basically does a 'replace-context'. But I was already doing that anyway.
It also serializes syntax properties that have symbol keys.
It works. My docs look nice again.
Neil T
_________________________________________________
For list-related administrative tasks:
http://lists.racket-lang.org/listinfo/dev
