Mac OS X 10.2.8 config

[Andrew Black <ab...@roguewave.com>]

Greetings all.

For those who are curious, attached is a copy of the config.h file 
generated when I ran the configuration process last night on my Mac OS X 
box.

I realize that the particular install is somewhat antiquated (I'm still 
on version 10.2.8, while the latest is 10.4.4 I believe), but I haven't 
gotten around to purchasing a more recent copy of the operating system.  
Geting a newer version has been something I've been planing to do, but 
just never gotten around to.  It also provides an interesting stress 
test to see just how well the library will perform in a somewhat lacking 
environment.

--Andrew Black

Re: Mac OS X 10.2.8 config

[Martin Sebor <se...@roguewave.com>]

Andrew Black wrote:
> Greetings all.
> 
> For those who are curious, attached is a copy of the config.h file 
> generated when I ran the configuration process last night on my Mac OS X 
> box.

Great, this might be useful for comparison with the latest release
of the OS.

Thanks!
Martin

Re: 21.string.append.cpp && 21.string.plus.equal.cpp

[Martin Sebor <se...@roguewave.com>]

Liviu Nicoara wrote:
> I attached a tar.gz archive containing the two tests mentioned in the
> subject line. The tests come from splitting the former append.cpp test
> and exercise lib.string.+= and lib.string.append.

Great! I committed it here:
http://svn.apache.org/viewcvs.cgi?rev=368132&view=rev

(Although I forgot your attribution -- sorry about that. Next
time I'll remember.)

Thanks a lot!
Martin

Re: 21.string.c.strings.cpp

[Martin Sebor <se...@roguewave.com>]

Liviu Nicoara wrote:
> I attached a tar.gz archive containing the test mentioned in the subject
> line (modified to use the new driver). The test comes from former
> c_strings.cpp test.

I'm actually in the process of changing this test in my local branch
so I'll need to merge your changes into mine before committing it.
Hopefully this weekend.

Thanks!
Martin

Re: 21.string.c.strings.cpp

[Martin Sebor <se...@roguewave.com>]

Liviu Nicoara wrote:
> I attached a tar.gz archive containing the test mentioned in the subject
> line (modified to use the new driver). The test comes from former
> c_strings.cpp test.

Thanks for the work on porting the test! Unfortunately, I'm afraid
the original test was very brittle and would tend to break on many
platforms due to slight deviations from the standard requirements.

To minimize the extent of the breakage it will be best to split the
test into several smaller programs, each exercising a single header.
In addition, instead of using the approach of taking the address of
the functions that are supposed to be declared in the header it would
be better to use templates to detect their absence (or non-conforming
signatures) without causing compilation errors.

I extracted the tests for the headers <cctype> and <cwctype> into
programs of their own and implemented the template solution mentioned
above here:

http://svn.apache.org/viewcvs.cgi/incubator/stdcxx/trunk/tests/strings/21.cctype.cpp?view=log
http://svn.apache.org/viewcvs.cgi/incubator/stdcxx/trunk/tests/strings/21.cwctype.cpp?view=log

It would be good to do the same for the rest of the headers that
the c_strings.cpp test exercises.

Thanks
Martin

21.string.c.strings.cpp

[Liviu Nicoara <ni...@roguewave.com>]

I attached a tar.gz archive containing the test mentioned in the subject
line (modified to use the new driver). The test comes from former
c_strings.cpp test.

Liviu

21.string.append.cpp && 21.string.plus.equal.cpp

[Liviu Nicoara <ni...@roguewave.com>]

I attached a tar.gz archive containing the two tests mentioned in the
subject line. The tests come from splitting the former append.cpp test
and exercise lib.string.+= and lib.string.append.

Liviu