[darcs-users] darcs patch: Data format and converter for "square" r... (and 4 more)

[Max Battcher]

A few brief, immediate responses:

On 05/29/2010 11:56 AM, Eric Kow wrote:
>> +  repoOrder a b = compare (reponame a) (reponame b)
>
> This can be written as compare `on` reponame
>
>    on :: (b ->  b ->  c) ->  (a ->  b) ->  a ->  a ->  c  Source
>    (*) `on` f = \x y ->  f x * f y.
>
> (The kind of thing one picks up through cultural osmosis)

Neat. That does seem useful. I figured there was probably a name for 
that, but `on` probably wouldn't have occurred to me.

> Refactor Run.hs to use updated tabulateRepo
> -------------------------------------------
>> -    case tabulate formatTimeResult res of
>> -      []  ->  return ()
>> -      [(_,t)] ->  echo_n $ TR.render id id id t
>> -      _   ->  error "Not expecting more than one table for a repo and its variants"
>> +    let tables = repoTables benchmarks res
>> +    if length tables == 1 then
>> +      echo_n $ TR.render id id id $ tabulateRepo formatTimeResult (head tables)
>> +      else error "Not expecting more than one table for a repo and its variants"
>
> Note that checking length and doing head is slightly less safe than just
> relying on pattern-matching with case (albeit perhaps more readable).
> It's unlikely that you'll break apart if length tables == 1 then foo
> (head tables), which is why I say "slightly"

If I do more Haskell I'll probably have to break myself of the head/tail 
over-reliance. (Prior to Haskell I did spend a few months hacking in 
Common Lisp where that pattern's so common there's a million names for 
it...) Luckily for me the list comprehension over-reliance from Python 
hacking seems to fit in reasonably well with Haskell. :)

> Also: do we never get the empty list?

I've got a feeling that it never happens. More than that, I'm pretty 
sure that Run.hs itself is deprecated and never actually used, but I 
wasn't sure enough to actually just go ahead and remove it. It seems to 
me to be an old "main" function from back before darcs-benchmark started 
using CmdArgs sub-commands, but I didn't take the time to do the 
darcs-fu to double check that against repository history. (Lazier and 
easier just to fix the compile errors by updating it. ;)

--
--Max Battcher--
http://worldmaker.net