Re: name:variable

On Nov 29, 2007, at 4:18 AM, Brandon Long wrote:

> Yeah, I don't think this is going to work.
>
> For instance, this expression won't work:
> <?cs var:url_escape(Request.Header.Accept-Encoding) ?>
>
> Even if we did the lookup in parse_expr instead, you'd get confusing
> results based on wether or not Request.Header.Accept-Encoding existed.

Agree, the patch was too simple and handled just one case. Just to  
prove that it is possible to do this general, enclosed a new patch  
which extend a simple branch in parse_tokens() to do a quick sub- 
scanning of input and do the HDF lookup. The new patch will handle  
compound statements such as <?cs var:url_escape(Request.Header.Accept- 
Encoding) ?>  etc.

I wont ask for acceptance of this patch because it just address my  
special case with HDF variables containing '-' in them, and should not  
be included in ClearSilver proper. However, if anything, what you may  
take away from the patch is to use a bit of table-driven lookup and  
save all those function calls to strpbrk() in csparse.c.

> Plus, HDF variable name lookups aren't the fastest things.  Also, this
> adds yet another place where vars have to exist at parse time.
>
> And I just don't think its that hard to not have vars with +/- in  
> them.
>
> CGI just munges Accept-Encoding to Accept_Encoding, the C++ wrapper I
> use at Google just munges it to AcceptEncoding, and I should probably
> just add that code to the cgi wrapper in ClearSilver.

Ah well, I think I have to revert to this then.

> Also, I'm pretty sure this works if you have to have these types of  
> var
> names:
>
> <?cs var:url_escape(Request.Header["Accept-Encoding"]) ?>

Yes, just verified, it works fine. Apropos, speaking about fine, I  
must say, now that I finally have taken a closer look at the  
clearsilver parse code, that it is very nice and clever indeed.

Best Regards