This program works, but what are these warnings all about? The program
compiles on g++ and bcc32 without complaint. From STL Tutorial and
Reference, Second Edition, Musser, Derge, Saini
 ex02-08.cpp, pp. 29-30

// Demonstrating the generic find algorithm with a deque
#include <iostream>
#include <cassert>
#include <deque>
#include <algorithm>  // For find
using namespace std;
template <typename Container>
Container make(const char s[])
{
  return Container(&s[0], &s[strlen(s)]);
}

int main()
{
  cout << "Demonstrating generic find algorithm with "
       << "a deque." << endl;
  deque<char> deque1 =
    make< deque<char> >("C++ is a better C");

  // Search for the first occurrence of the letter e:
  deque<char>::iterator
    where = find(deque1.begin(), deque1.end(), 'e');
  assert (*where == 'e' && *(where + 1) == 't');
  cout << " --- Ok." << endl;
  return 0;
}:

 C:\dm\bin\sc -Ae -C -WA -S -5 -a8 -o ex02-08.exe
ex02-08.cpp -I\dm\stlport\stlport
ex02-08.cpp(27) : Warning 12: variable 'char ** __cur_node' used before set
ex02-08.cpp(27) : Warning 12: variable 'char ** __cur' used before set
ex02-08.cpp(27) : Warning 12: variable 'char ** __cur' used before set
link ex02-08,ex02-08,,user32+kernel32/noi;

I've had some similar ones from STLPort. I think it's because some STLPort functions throw as the last action within function bodies, so there's no return statement.

Oh, no, sorry. That was my errors. Doesn't seem to be the explanation for yours.

I'd like to put some effort into the STLPort, if no-one else is - if they are then am more than happy to leave it - but time's not going to be friendly for a while.

Is there someone that has taken the STLPort under their wing? (Is that
Christof?)



"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b72m69$1oin$1@digitaldaemon.com...
> This program works, but what are these warnings all about? The program
> compiles on g++ and bcc32 without complaint. From STL Tutorial and
> Reference, Second Edition, Musser, Derge, Saini
>  ex02-08.cpp, pp. 29-30
>
> // Demonstrating the generic find algorithm with a deque
> #include <iostream>
> #include <cassert>
> #include <deque>
> #include <algorithm>  // For find
> using namespace std;
> template <typename Container>
> Container make(const char s[])
> {
>   return Container(&s[0], &s[strlen(s)]);
> }
>
> int main()
> {
>   cout << "Demonstrating generic find algorithm with "
>        << "a deque." << endl;
>   deque<char> deque1 =
>     make< deque<char> >("C++ is a better C");
>
>   // Search for the first occurrence of the letter e:
>   deque<char>::iterator
>     where = find(deque1.begin(), deque1.end(), 'e');
>   assert (*where == 'e' && *(where + 1) == 't');
>   cout << " --- Ok." << endl;
>   return 0;
> }:
>
>  C:\dm\bin\sc -Ae -C -WA -S -5 -a8 -o ex02-08.exe
> ex02-08.cpp -I\dm\stlport\stlport
> ex02-08.cpp(27) : Warning 12: variable 'char ** __cur_node' used before
set
> ex02-08.cpp(27) : Warning 12: variable 'char ** __cur' used before set
> ex02-08.cpp(27) : Warning 12: variable 'char ** __cur' used before set
> link ex02-08,ex02-08,,user32+kernel32/noi;
>
>
>

In article <b72obm$1ptq$1@digitaldaemon.com>, Matthew Wilson says...

>I'd like to put some effort into the STLPort, if no-one else is - if they are then am more than happy to leave it - but time's not going to be friendly for a while.
>
>Is there someone that has taken the STLPort under their wing? (Is that
>Christof?)

While I can't speak for Christof, I can say that if I have serious problems with stlport, Christof is the one I'm going to go begging to.

As for the warnings, I mentioned in a much earlier thread that these were warnings that appeared whenever a release build was made using deque (and possibly a few others). I also added that they were annoying, but probably harmless since no one has tracked them down. Since the thread has appeared again, they are no longer "harmless" as they are wasting time.. so:

void fn(int i) { }

int main() {
int i;
try {

} catch (...) {
fn(i);
}
}
// Warning 12: variable 'int  i' used before set

This is a design point made in STL for efficiency. __cur_node is assigned inside the try block, and exceptions can only be thrown after the assignment.

I don't know what Walter wants to do with this. Warnings are generated for optimized builds, but not for non optimized builds.

Richard

"Richard Grant" <fractal@clark.net> wrote in message news:b74gfm$2uo7$1@digitaldaemon.com...
> In article <b72obm$1ptq$1@digitaldaemon.com>, Matthew Wilson says...
>

 Since the thread has appeared
> again, they are no longer "harmless" as they are wasting time.. so:
>
> void fn(int i) { }
>
> int main() {
> int i;
> try {
>
> } catch (...) {
> fn(i);
> }
> }
> // Warning 12: variable 'int  i' used before set
>
>
> Richard
>

I discovered that the warning messages are caused by the space between the -o flag and the output file name in the command line:

C:\dm\bin\sc -Ae -C -WA -S -5 -a8 -o ex02-08.exe ex02-08.cpp -I\dm\stlport\stlport

Take out the space and the warnings go away.
I will not waste any more time of my time.
Jim J.

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b74u7v$5rk$1@digitaldaemon.com...
> I discovered that the warning messages are caused by the space between the -o flag and the output file name in the command line:
>
> C:\dm\bin\sc -Ae -C -WA -S -5 -a8 -o ex02-08.exe ex02-08.cpp -I\dm\stlport\stlport
>
> Take out the space and the warnings go away.
> I will not waste any more time of my time.
> Jim J.

-o by itself means "optimize". -ofilename means write the output to filename.

"Walter" <walter@digitalmars.com> wrote in message news:b753dc$9bu$2@digitaldaemon.com...
>
> "Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b74u7v$5rk$1@digitaldaemon.com...
> > I discovered that the warning messages are caused by the space between the -o flag and the output file name in the command line:
> >
> > C:\dm\bin\sc -Ae -C -WA -S -5 -a8 -o ex02-08.exe ex02-08.cpp -I\dm\stlport\stlport
> >
> > Take out the space and the warnings go away.
> > I will not waste any more time of my time.
> > Jim J.
>
> -o by itself means "optimize". -ofilename means write the output to filename.
>
>
I am using a makefile written for the GNU compiler, that I adapted for dmc.
g++ has an option -o <file> "place output into file", (with space). That's
the way the makefile is written, -o $@  -- with a space -- and it compiles
with g++. I think that -ofilename  and -o optimizer flag (whatever they are)
is a very fine distinction. Probably not one person in a thousand coming
over from gcc would be diligent enough to catch the difference.
However, all this is beside the point. Digital Mars is now relegated to file
13. Apparently everyone is happy working around the bugs and gratuitous
warnings. So there is no point in posting them anymore.

"Richard Grant" <fractal@clark.net> wrote in message news:b74gfm$2uo7$1@digitaldaemon.com...
> In article <b72obm$1ptq$1@digitaldaemon.com>, Matthew Wilson says...
>
 Since the thread has appeared
> again, they are no longer "harmless" as they are wasting time.. so:
>
> void fn(int i) { }
>
> int main() {
> int i;
> try {
>
> } catch (...) {
> fn(i);
> }
> }
> // Warning 12: variable 'int  i' used before set
>
> This is a design point made in STL for efficiency. __cur_node is assigned
inside
> the try block, and exceptions can only be thrown after the assignment.
>
> Richard
>
I am missing something here? What are you saying? That I should put a
try/catch sequence in my source code to suppress a compiler warning? Or did
I overlook another smiley?
James

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b758nl$cma$1@digitaldaemon.com...
> I am using a makefile written for the GNU compiler, that I adapted for
dmc.
> g++ has an option -o <file> "place output into file", (with space). That's the way the makefile is written, -o $@  -- with a space -- and it compiles with g++. I think that -ofilename  and -o optimizer flag (whatever they
are)
> is a very fine distinction. Probably not one person in a thousand coming over from gcc would be diligent enough to catch the difference. However, all this is beside the point. Digital Mars is now relegated to
file
> 13. Apparently everyone is happy working around the bugs and gratuitous warnings. So there is no point in posting them anymore.

If I changed the way the command line worked, it would break innumerable existing makefiles. It's not really an option to do that. Every compiler has odd command line syntax in one way or another, there's no standard, it's all in what you're used to. The existing syntax was not really designed to be that way, it is an accumulation of 20 years of adding features while retaining backward compatibility.

In the unix world, it's convention to put a space between a switch and its argument. In the DOS/Win32 world, the convention is to not have a space between. Neither is wrong or right, it's just convention.

What I do with projects for multiple compilers is just go ahead and do a unique makefile for each compiler.