Re: LSR categories

[Han-Wen Nienhuys]

Graham Percival escreveu:
>> Agree! Often such division into different categories don't help at
>> all, they just give you more places to search from.
> 
> Other than one example (scheme programming), I have no objection to this
> -- but please don't say "text, lyrics, articulation etc".  I'd like to
> discuss exact proposals.
> 
> For example, are you proposing categories of
> text
> lyrics
> articulations
> dynamics
> staff
> beaming
> key and time signatures
> piano stuff
> percussion
> ancient
> titles/presentation
> 
> or are you suggesting
> text
> lyrics
> notation
> titles/presentation

Why do you use exclusive categories?  Isn't it better to devise some
sort of tagging scheme, eg.

  \header {
     subjects = "lyrics, scheme, piano, titles"
  }

and either

  - produce 1 big page with 

       Tag:lyrics
       Tag:scheme
       Tag:piano
         ....

    prefixed to each example, so you can easily search.

 - produce a page per tag: a page for all lyrics examples, a page for
   all piano stuff etc.

   This is a bit unnatural, though, so it would be better if we could
   find some lightweight method to dynamically generate those
   pages. It be possible with javascript, but I'm no export at
   JavaScript.

> The result will look very different in these two proposals!  In the
> first, we average 5-20 snippets per category; in the second, we have
> 10-100 snippets per category.  Once we include the snippets currently on
> the website, this would reach 15-300 snippets per category.
> 
> There's obviously a middle ground that we need to reach, but I think the
> best way to find this is to discuss exact proposals.
> 
> 
>> (preferably including Regression Tests and maybe even all examples
>> from the manual) 
> 
> Only if we had an infinite supply of data entry volunteers, I'm afraid.
>  My preferred order is
> - current LSR snippets
> - start a big push on -user to get people adding their own snippets
> - current "examples" and "tricks and tips"
> - current regression tests
> 
> Without any new snippets from users and without the regression tests,
> that's about 250 snippets to examine.  There's just under 500 regression
> tests; not all are useful to add, but somebody still needs to check them
> all and decide.

I recommend against using the regression test as a source of
interesting snippets. The collection is growing quite rapidly (I add
one every time I fix a bug), and is not designed to be interesting for
users. In fact, I'm deconstructing most of the longer ones, when I run
into them so they are more usable for debugging.


-- 

Han-Wen Nienhuys - address@hidden - http://www.xs4all.nl/~hanwen

LilyPond Software Design
 -- Code for Music Notation
http://www.lilypond-design.com