16:59 <jugglinmike> I want to create a "table of contents" page that lists links to all the pages in a given directory. I think that to do this, I would have to create a global array (say, `classes`) and push items into it in the `preprocess` block based on their `identifier` property (i.e. `if item.identifier[/^/training/classes]` ). Then, the "table of contents" layout could iterate over the `classes` array. Is that the best way to accompli

20:22 <dkm> jugglinmike: I did that like this https://github.com/hazybluedot/www-source/blob/master/content/reference/index.slim

20:23 <dkm> In my case, I prefaced pages I wanted to be in the index with '_', although I could have also filtered the list of items to exclude the index itself, I think

20:23 <jugglinmike> I see dkm--thanks!

20:24 <jugglinmike> I'm trying to remove the logic from my layouts, so I may go with the `preprocess` approach

20:24 <dkm> yeah, I'm not necessarily in love with my implementation for that reason

20:25 <dkm> although in this case, I think of it less as "logic" and more of "generation of content at compile time"

20:25 <dkm> a slippery slope, to be sure!

20:27 <dkm> one could also have philisophical issues with creating a global array of items that really wasn't logically global

20:30 <bobthecow> jugglinmike: i tend to move logic like that into helpers in lib.

20:30 <bobthecow> i have an `articles` method, a `projects` method, etc.

20:32 <jugglinmike> dkm: I think we could debate that, but unfortunately, I'm preoccupied at the moment :P

20:32 <dkm> bobthecow: I like that, then jugglinmike could create a method like #get_index_for(item) or something

20:32 <bobthecow> yeah.

20:32 <dkm> I might change my implementation to do that, thanks for the idea!

20:32 <bobthecow> except the "get" part isn't very ruby.

20:33 <bobthecow> just index_for

20:33 <dkm> ah, ok

20:39 <jugglinmike> bobthecow dkm: This function sounds interesting, but I'm not sure I understand. Which item would I pass to `index_for` ? The "table of contents" item? And I guess it would return an array of all the items that belonged?

20:39 <dkm> jugglinmike: perpahs it would be an 'item' but a path

20:39 <dkm> #index_for(path)

20:39 <dkm> which would return an array of items

20:40 <dkm> the index page would still have to iterate over a list of items to generate an index item

20:40 <dkm> but I suppose you could put *that* in a pre-process step

21:14 <dkm> hmm, I wonder if it makes sense to make this a DataSource?

21:17 <dkm> IndexDataSource#up would walk through the directories in /content, creating an index page for each subdirectory, IndexDataSource#items would then return the array of index page items

21:17 <dkm> you'd probably want a way to exclude a subset of subdirectories from the indexer

21:18 <bobthecow> dkm: iterating over a list of items in memory is *really* fast.

21:18 <bobthecow> that'sunlikely to be a bottleneck for you.

21:18 <dkm> bobthecow: I'm just thinking in terms of what would be most nanocic

21:19 <bobthecow> a helper method.

21:19 <dkm> but you are in fact generating nanoc items

21:19 <bobthecow> https://github.com/nanoc/nanoc/blob/master/lib/nanoc/helpers/blogging.rb#L30-L37

21:19 <bobthecow> oh, you're creating index items that don't already exist?

21:19 <dkm> right, the blogger helper is just returning a subset of already generated items

21:20 <bobthecow> i thought you were just finding things that belong in an index.

21:20 <bobthecow> yeah, then create items in preprocess.

21:20 <dkm> bobthecow: that's how I understood jugglinmike's request

21:20 <bobthecow> it could be a data source, but that's a bit of a weird way to do it.

21:21 <dkm> ok, yeah, I kind of think of data sources as generating items from content outside of nanoc's scope

21:21 <dkm> or... no

21:21 <dkm> that wasn't really a good explaination of how I view data sources

21:22 <dkm> I think it kind of depends how we think of generating an index page

21:22 <jugglinmike> bobthecow: you're right there

21:22 <bobthecow> i'd do it like the way the "tag page" example does it.

21:22 <dkm> are we walking a directory structure on the file system, or are we iterating over a list of nanoc identifiers

21:24 <bobthecow> dkm: everything in nanoc is items.

21:24 <bobthecow> some of 'em just happen to be persisted to the filesystem :)

21:24 <dkm> bobthecow: yeah, I think the tag page is idea is very close to what the index page would be

22:10 <dkm> bobthecow, jugglinmike: my new implementation https://github.com/hazybluedot/www-source/blob/master/lib/index_helper.rb

22:11 <dkm> I still created a layout for index_page because I don't like the idea of embeding html in the helper function

22:11 <dkm> oh woops, and I'm still hard-coding the identifier

22:11 <dkm> need to derive it from the path

22:17 <dkm> updated

22:18 <dkm> see line 26 in my Rules for where I use it in the preprocess block

22:18 <dkm> sorry, line 29