C++ sample source to test compiler compliancy

[spipyeah]

Is there a collection of C++ source files that can be used to test
compiler compliancy by using really extreme language cases, making
sure the standard library is complete, and other such things?

 

[Attila Feher]

Is there a collection of C++ source files that can be used to test
compiler compliancy by using really extreme language cases, making
sure the standard library is complete, and other such things?

If you want to know about a specific compiler then I think at Boost you can
find out a lot.

 

[tom_usenet]

Is there a collection of C++ source files that can be used to test
compiler compliancy by using really extreme language cases, making
sure the standard library is complete, and other such things?

http://www.plumhall.com/suites.html
http://www.peren.com/pages/cppvs.htm
http://www.dinkumware.com/cproofer.html (library only)

also:
http://www.cuj.com/documents/s=8027/cuj0104sutter/sutter.htm

I'm not aware of anything free that attempts to check conformance
properly.

Tom

 

[tom_usenet]

If you want to know about a specific compiler then I think at Boost you can
find out a lot.

The problem is that boost makes extensive use of workarounds to make
code work for non-standard compilers. The percentage of passed boost
regression tests is far more an indication of the amount of work that
has been put into porting boost to that compiler than of the
conformance of that compiler.

e.g. g++ and Metroworks both score "better" than Comeau C++.

Tom

 

[Attila Feher]

The problem is that boost makes extensive use of workarounds to make
code work for non-standard compilers. The percentage of passed boost
regression tests is far more an indication of the amount of work that
has been put into porting boost to that compiler than of the
conformance of that compiler.

e.g. g++ and Metroworks both score "better" than Comeau C++.

That is why I did not say: look at the Boost compatibility charts, but that
"at Boost you can find out a lot." What I meant is that if he has a
specific compiler in mind he can look at what workarounds are there for it
in Boost, read (possibly ask) on the mailing list etc.

 

[Greg Comeau]

The problem is that boost makes extensive use of workarounds to make
code work for non-standard compilers. The percentage of passed boost
regression tests is far more an indication of the amount of work that
has been put into porting boost to that compiler than of the
conformance of that compiler.

e.g. g++ and Metroworks both score "better" than Comeau C++.

Right, Boost does not provide a measure of compliance to
Standard C++. I suspect the OP is looking for some test
suites, though "extreme language cases" may be asking for more.

 

[Greg Comeau]

http://www.plumhall.com/suites.html
http://www.peren.com/pages/cppvs.htm
http://www.dinkumware.com/cproofer.html (library only)

also:
http://www.cuj.com/documents/s=8027/cuj0104sutter/sutter.htm

I'm not aware of anything free that attempts to check conformance
properly.

I don't know the critereas used, but I understand there
is also an upcoming article in DDJ due shortly too.

 

[spipyeah]

Thank you all, I'll be checking all those resources, and keep an eye on DDJ too.

 

[spipyeah]

Creg, if I may ask, in your experience, how does GNU g++ usually fare?

 

[David B. Held]

Creg, if I may ask, in your experience, how does GNU g++
usually fare?

In my opinion, gcc 3.3 is outstanding. It might have bugs, but
I haven't run across one yet!

Dave

 

[White Wolf]

In my opinion, gcc 3.3 is outstanding. It might have bugs, but
I haven't run across one yet!

3.3 or 3.3.1?

And I guess you mean that at the features claimed to be supported. I guess
two phase name lookup is not there yet and ordering... but I could not find
their compliance statement.

 

[llewelly]

3.3 or 3.3.1?

And I guess you mean that at the features claimed to be supported. I guess
two phase name lookup is not there yet and ordering... but I could not find
their compliance statement.

It is my impression that gcc 3.4 will have two-phase lookup. gcc
3.3.x does not. 3.3.1 is only a bugfix release on 3.3 .