RE: template template parameters (was Re: [boost] Re: Special fun ctions, octonions, quaternions in the CVS)

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-06-26 10:42:40

Aleksey Gurtovoy wrote:

 > Beman Dawes wrote:
 >> But in another message someone claimed that template template
 >> parameter replacements were not just broken compiler
 >> workarounds, but actually better that template template parameters.
 >> Could someone explain the rationale behind that assertion?
 >
 >One of the reasons is described here -
 >http://groups.yahoo.com/group/boost/message/8454. I've heard, however,
 >Dietmar saying that there is a preliminary agreement to fix this so the
 >second example in the cited message would work too. Don't know the 
details,
 >though.

Thanks.  I'd missed that.

I can see that problem (default argument causing parameter list mismatch 
error) as a reason to use the helper approach in some cases.  Say when you 
don't expect to have total ability to set exact requirements for classes 
used to instantiate, perhaps because they may come from third-party 
libraries.

But for cases like policy classes which are be written specifically to 
model a generic concept, it wouldn't be a major problem would it?  Since 
writing the model is slightly simpler for use with template templates, 
wouldn't that be preferred?

--Beman