markr@and.cs.liv.ac.uk (06/17/91)
I started a C++ project that needed many generic classes (sets, lists etc.) and
decided to use templates. Until we get a compiler that supports them I am
using the cpp from ti.
I notice recently there has been a lot of talk of templates on this group. I
don't know much about the situation, and would like to ask:-
* are templates part of a standard for C++, or just a proposal?
* do any compilers support them yet?
* when will the AT&T compiler support them? 2.3, 3.0, 16.8 or what?
* when they are supported, how efficient will they be in terms of
avoiding code duplication? Will it be necessary to define a general
base class and use templates for type-specific derived classes?
(eg. template <class T> class Set<T> : public GenericSet {...})
* how does use of templates with the ti macro package differ from the
standard/proposal for the language?
--
Mark Rivers
CompSci Liverpool Uni UKsteve@taumet.com (Stephen Clamage) (06/19/91)
markr@and.cs.liv.ac.uk writes: >I notice recently there has been a lot of talk of templates on this group. I >don't know much about the situation, and would like to ask:- >* are templates part of a standard for C++, or just a proposal? There is no standard for C++ yet. E&S (the ARM) describes templates, and is the model the ANSI C++ committee is using. The final standard for C++ will include templates, and will look much like what is in E&S. Templates in their final form may or may not be identical to what is now in E&S. >* do any compilers support them yet? No released commercial compilers that I know of. >* when they are supported, how efficient will they be in terms of > avoiding code duplication? Will it be necessary to define a general > base class and use templates for type-specific derived classes? > (eg. template <class T> class Set<T> : public GenericSet {...}) Read E&S for a description of how they work. >* how does use of templates with the ti macro package differ from the > standard/proposal for the language? The TI implementation is not a full template implementation. The whole job requires more than can be done by a preprocessor. -- Steve Clamage, TauMetric Corp, steve@taumet.com
jimad@microsoft.UUCP (Jim ADCOCK) (06/19/91)
In article <1991Jun16.173916.143@and.cs.liv.ac.uk| markr@and.cs.liv.ac.uk writes: |I notice recently there has been a lot of talk of templates on this group. I |don't know much about the situation, and would like to ask:- | |* are templates part of a standard for C++, or just a proposal? Last I had heard, the ANSI C++ committee had voted templates "in" the language spec -- and then immediately turned around and started arguing about the exact syntax that should be used for templates. Turns out I guess that using '<' and '>' as template brackets leads to some pretty ambiguous parsing [probably not surprising considering that '<' and '>' have radically different meanings throughout the rest of the language] Still, to date, I think the description in Ellis and Stroustrup's ARM -- "The Annotated C++ Reference Manual" -- is the best reference on templates. [ Check out the updated Lippman when it comes out :-]
sarima@tdatirv.UUCP (Stanley Friesen) (06/19/91)
In article <1991Jun16.173916.143@and.cs.liv.ac.uk> markr@and.cs.liv.ac.uk writes: > >I notice recently there has been a lot of talk of templates on this group. I >don't know much about the situation, and would like to ask:- > >* are templates part of a standard for C++, or just a proposal? They are part of the proposed standard. (The committee is firm on this). >* do any compilers support them yet? I am not sure, the latest version of cfront from ATT has added one of exception handling or templates (but I do not remember which). Also, I have heard that at least one specialty C++ compiler has an implementation. >* when they are supported, how efficient will they be in terms of > avoiding code duplication? Will it be necessary to define a general > base class and use templates for type-specific derived classes? > (eg. template <class T> class Set<T> : public GenericSet {...}) This will be implementation dependent. As I understand it most currently proposed implementations rather tend towards duplicate code. -- --------------- uunet!tdatirv!sarima (Stanley Friesen)
wlp@calmasd.Prime.COM (Walter L. Peterson, Jr.) (06/20/91)
In article <49@tdatirv.UUCP> sarima@tdatirv.UUCP (Stanley Friesen) writes: >In article <1991Jun16.173916.143@and.cs.liv.ac.uk> markr@and.cs.liv.ac.uk writes: >>* do any compilers support them yet? > >I am not sure, the latest version of cfront from ATT has added one of >exception handling or templates (but I do not remember which). >Also, I have heard that at least one specialty C++ compiler has an >implementation. > The C++ compiler that comes with the ObjectStore OODBS from Object Design has implemented templates. I have not yet used this myself, but AT&T did license this implementation of templates from Object Design for its own use, so that is something of an endorsement. ObjectStore is made and sold by: Object Design One New England Executive Park Burlington, MA 01803 Phone: (617) 270-9797 FAX: (617) 270-3509 -- "Exploring the Consensual Hallucination of Cyberspace" Walter L. Peterson, Jr. Internet : wlp@calmasd.Prime.COM | CompuServe : 70441,3144 | GEnie: WPETERSON5 "Insert generic disclaimer here"
ianhogg@cs.umn.edu (Ian J. Hogg) (06/21/91)
In article <773@taumet.com> steve@taumet.com (Stephen Clamage) writes: >markr@and.cs.liv.ac.uk writes: > >There is no standard for C++ yet. E&S (the ARM) describes templates, >and is the model the ANSI C++ committee is using. The final standard >for C++ will include templates, and will look much like what is in E&S. >Templates in their final form may or may not be identical to what is >now in E&S. > >>* do any compilers support them yet? > >No released commercial compilers that I know of. > The version of Cfront you get with ObjectStore has templates. AT&T has licensed the code from ObjectDesign. -- =============================================================================== Ian Hogg ianhogg@cs.umn.edu (612) 225-1401