On Sun, Jun 21, 2009 at 10:29 PM, Don Stewart wrote:

gwern0:

...

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512

So I mentioned a while ago I meant to go over XMC using the 'sim' tool (with an invocation like sim_mira `find . -name "*.hs"` - I discovered recently that it had Miranda support, which means quasi-Haskell support). But the results are littered with comments.

Particularly obnoxious are the module headers, in part because they all seem to use '--' comments even if they cover dozens of lines. Would anyone object if I were to edit those to use {- -} style (and push those edits without review)?

Yep. Let's not change the style. But change the tool to support the style.

I'm not familiar with 'sim'. What does it do?

-- Don

As in my previous email in April on the topic: sim is a tool for discovering textual repetition http://www.haskell.org/pipermail/xmonad/2009-April/007722.html It works reasonably well, but the output is much harder to read than it needs to be because of the module headers. As you can see, mailman mangled my email, so I'm forwarding the original to you & Spencer (as the only 2 to reply). -- gwern

On Tue, Jun 30, 2009 at 9:05 PM, Adam Vogt wrote:

You could strip comments with something like this. The results might be slightly better, depending on how well sim works. Haskell-src-exts is improving, but it still fails on a couple modules in contrib.

#!/usr/bin/env runghc

...

import System.Environment import Language.Haskell.Exts import Control.Monad

...

main = mapM_ (putStrLn . prettyPrint . fromParseResult <=< parseFile) =<< getArgs

That's useful, but I think it's not as good as changing the module headers. As you say, that will choke on some modules; it's an extra step; and by removing comments entirely, one can no longer investigate any duplication in *comments* as well as code (I'm sure copy-pasta of comments exists, even if only in the module headers.) And I don't entirely understand the reticence to use multi-line comments. Weren't they intended for exactly these sorts of situations, massive comment blocks spanning entire pages? How exactly are a couple hundred prefixed '--'s superior to a {- -}? -- gwern

On Fri, Dec 18, 2009 at 04:33:56PM -0500, Gwern Branwen wrote:

So with a potentially long wait until the next major release (that is, not the quick fixes for X11/GHC), does anyone mind if I were to make these header changes?

As I've said already, please don't change the module headers. If sim is broken, fix sim. Cheers, Spencer Janssen

Gwern Branwen writes:

As I said already: sim is not broken. Our conventions are broken. The headers have massive duplication, and sim is reporting the duplication exactly as it should.

Just because we have reptition doesn't mean our conventions are "broken"; inefficient maybe, but I wouldn't even really consider it to be "duplication". Besides, M-q in Emacs helps with editing... ;-) -- Ivan Lazar Miljenovic Ivan.Miljenovic@gmail.com IvanMiljenovic.wordpress.com