language.c: support for statements in MonadTrav

Benedikt Huber benedikt.huber at gmail.com
Fri Feb 4 16:46:13 GMT 2011


On 04.02.2011, at 17:05, Alexander Bernauer wrote:

> Hi
> [..]
>> Ah, sorry. What I had in mind is that visitors are not necessary if
>> you can use generic programming
>> [http://d.hatena.ne.jp/syd_syd/20080813#p1].
>
> I can't read nay Japanese, can you?
Hi,
No :) But if you know a little bit about Data.Generics the google- 
translated
webpage gives you a very good idea what it means to manipulate
syntax trees using 'everywhere' and friends.
>
> But I googled a bit and found the Lämmel & Jones papers [1] and I  
> guess
> you refer to that.
correct.
>
> I am pretty new to Haskell so I am not completely sure about the
> following. But I think generic programming depends on run-time type
> checks which could be expensive at runtime, couldn't it?
They do depend on run-time type checks, obviously; I have no idea
how 'expensive' generic programming is, though. I think Neil Mitchell
did performance measurements for his Uniplate (http://hackage.haskell.org/package/uniplate 
)
and a few other frameworks, but I can't find the page right now.

> On the other hand I think a generic visitor implementation also  
> requires
> having the boilerplate code for traversal only once. So, what  
> advantages do
> you see for using generic programming?
I think visitor frameworks and generic programming are related concepts.
Renaming all variables by writing
 > let ast' = everywhere (mkT renameVarExpr) ast
looks elegant and 'functional' to me, but maybe you can achieve
the same thing using visitors? A (in this cass imho less important)  
benefit
of generic programming is that the library author does not have to  
write any code.
But you definitely benefit from an language-c specific visitor or  
generic
programming framework; you know that all visited elements are CNodes,
for example; or (as I said before) provide symbol table and type  
information.

>> But on the raw C AST, generic programming is not trivial, because
>> you need to have symbol tables and type information at hand to do
>> something useful. For example, when building a callgraph, you need
>> to know the current state of the symbol table at each call site,
>> because a local definition might shadow a global one.
>
> As far as I know there are only local function declarations but no
> definitions. Furthermore, local function declarations always refer to
> global function definitions. This means, you can only change the  
> type of
> a function locally but you can not shadow a global function.
>
> So, collecting the call graph should be easy. Simply find all function
> calls in all function definitions and use the function name as a
> reference.  Did I miss anything?
Yes, you did :)
 > int f(int (*g)(void)) {
 >    return g();
 > }

>
>> CIL does a lot of preprocessing, which greatly simplifies the C AST.
>> If this is an option for you, maybe [1] would be interesting to you.
>
> Hmm, interesting. But this seems to be a bit incomplete, doesn't it?
>
>> Otherwise, you will benefit from Language.C exposing all the
>> cruelities of the C Syntax :)
>
> Not the syntax but the semantics. Nevertheless, it's cruel, I know :-/
>
> I think I will give the visitor framework a try.

Ok fine. I hope you'll make good progress :)

cheers, benedikt

p.s.: I'm on holiday for 3 weeks now, so replies will be delayed even
more than usual ;)

>
> best
>
> Alex
>
> PS: I am subscribed to the list now, so you don't have to CC me any
> more. Thanks.
>
> [1] http://research.microsoft.com/en-us/um/people/simonpj/papers/hmap/




More information about the Language-c mailing list