The in_file.bad() instruction in the code below does not return true if a
bad file name is entered on the command line. The cerr message and the
return are skipped. The two commented lines return true, and the error
message and the return statement execute for both of them. I am using the
stlport library with dmc.
According to N. Josuttis, C++ Standard Library, p.598, bad() is supposed to
return true if  a fatal error has occurred (badbit is set). I have compiled
this program with dmc and another compiler with identical results.

    std::ifstream in_file(argv[1]);
//    if (in_file.fail())
//    if (!in_file)
    if (in_file.bad())
    {
         std::cerr << "Cannot open " << argv[1] << '\n';
         return -1;
    }

I can find bad() in dm\include\iostream.h and some dll's in dm\bin
iostream.h says this:
inline int      ios::fail() { return state & (failbit | badbit |
hardfail); }
inline int      ios::bad() { return state & (badbit | hardfail); }

Am I correct in concluding that trying to open a non-existing input file does not turn on either the badbit or the hardfail, but only the failbit? Jim W. J.

Try the version of <iostream> in \dm\stlport\stlport, the older <iostream.h> in \dm\include is obsolete (but included for backwards compabitility).

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b4harj$2fmg$1@digitaldaemon.com...
> The in_file.bad() instruction in the code below does not return true if a
> bad file name is entered on the command line. The cerr message and the
> return are skipped. The two commented lines return true, and the error
> message and the return statement execute for both of them. I am using the
> stlport library with dmc.
> According to N. Josuttis, C++ Standard Library, p.598, bad() is supposed
to
> return true if  a fatal error has occurred (badbit is set). I have
compiled
> this program with dmc and another compiler with identical results.
>
>     std::ifstream in_file(argv[1]);
> //    if (in_file.fail())
> //    if (!in_file)
>     if (in_file.bad())
>     {
>          std::cerr << "Cannot open " << argv[1] << '\n';
>          return -1;
>     }
>
> I can find bad() in dm\include\iostream.h and some dll's in dm\bin
> iostream.h says this:
> inline int      ios::fail() { return state & (failbit | badbit |
> hardfail); }
> inline int      ios::bad() { return state & (badbit | hardfail); }
>
> Am I correct in concluding that trying to open a non-existing input file does not turn on either the badbit or the hardfail, but only the failbit? Jim W. J.
>
>
>

Walter,
I believe I am using <iostream> in \dm\stlport\stlport. I #include
<iostream> in the program. I have
also, per instructions in the dm\stlport\readme.txt file, replaced a line in
sc.ini:

Quote : To use STLport without iostreams, simply add \dm\stlport\stlport to
the
INCLUDE search path before \dm\include. Or, modify the INCLUDE entry
in \dm\bin\sc.ini to be:

INCLUDE="%@P%\..\stlport\stlport";"%@P%\..\include";"%@P%\..\mfc\include";%I
NCLUDE%
Unquote:
Making a wild guess, in sc.ini, I have replaced
INCLUDE="%@P%\..\include";"%@P%\..\mfc\include";"%@P%\..\stl";%INCLUDE%
with:
INCLUDE="%@P%\..\stlport\stlport";"%@P%\..\include";"%@P%\..\mfc\include";%I
NCLUDE%

This change works (I do not have to say -I\dm\stlport\stlport on the command
line anymore), but I am confused by the phrase "To use STLport without
iostreams, . . ." when later in the readme.txt file it says:
Quote: To compile a program using STLport's <iostreams> with the static
library:
 sc hello -I\dm\stlport\stlport
Unquote. Which is it? The statements are contradictory. Are these two
statements referring to two different iostreams, STLport's and some other
one, or do they both refer to the STLport iostream? What is the "static
library"?

Following Christof's instructions I have:
(built) the STLport libraries and dll's with: smake -f dm.mak

bye, bye,
Jim J.

"Walter" <walter@digitalmars.com> wrote in message news:b4jotn$bfo$2@digitaldaemon.com...
> Try the version of <iostream> in \dm\stlport\stlport, the older
<iostream.h>
> in \dm\include is obsolete (but included for backwards compabitility).
>
> "Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b4harj$2fmg$1@digitaldaemon.com...
> > The in_file.bad() instruction in the code below does not return true if
a
> > bad file name is entered on the command line. The cerr message and the return are skipped. The two commented lines return true, and the error message and the return statement execute for both of them. I am using
the
> > stlport library with dmc.
> > According to N. Josuttis, C++ Standard Library, p.598, bad() is supposed
> to
> > return true if  a fatal error has occurred (badbit is set). I have
> compiled
> > this program with dmc and another compiler with identical results.
> >
> >     std::ifstream in_file(argv[1]);
> > //    if (in_file.fail())
> > //    if (!in_file)
> >     if (in_file.bad())
> >     {
> >          std::cerr << "Cannot open " << argv[1] << '\n';
> >          return -1;
> >     }
> >
> > I can find bad() in dm\include\iostream.h and some dll's in dm\bin
> > iostream.h says this:
> > inline int      ios::fail() { return state & (failbit | badbit |
> > hardfail); }
> > inline int      ios::bad() { return state & (badbit | hardfail); }
> >
> > Am I correct in concluding that trying to open a non-existing input file does not turn on either the badbit or the hardfail, but only the
failbit?
> > Jim W. J.

stlport can be used with and without iostreams. To use the iostreams, you'll need to link with the stlport static library. If using stlport without iostreams, you don't need to link with that library. Sorry about the confusion. -Walter

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b4l1p6$12bc$1@digitaldaemon.com...
> Walter,
> I believe I am using <iostream> in \dm\stlport\stlport. I #include
> <iostream> in the program. I have
> also, per instructions in the dm\stlport\readme.txt file, replaced a line
in
> sc.ini:
>
> Quote : To use STLport without iostreams, simply add \dm\stlport\stlport
to
> the
> INCLUDE search path before \dm\include. Or, modify the INCLUDE entry
> in \dm\bin\sc.ini to be:
>
>
INCLUDE="%@P%\..\stlport\stlport";"%@P%\..\include";"%@P%\..\mfc\include";%I
> NCLUDE%
> Unquote:
> Making a wild guess, in sc.ini, I have replaced
> INCLUDE="%@P%\..\include";"%@P%\..\mfc\include";"%@P%\..\stl";%INCLUDE%
> with:
>
INCLUDE="%@P%\..\stlport\stlport";"%@P%\..\include";"%@P%\..\mfc\include";%I
> NCLUDE%
>
> This change works (I do not have to say -I\dm\stlport\stlport on the
command
> line anymore), but I am confused by the phrase "To use STLport without
> iostreams, . . ." when later in the readme.txt file it says:
> Quote: To compile a program using STLport's <iostreams> with the static
> library:
>  sc hello -I\dm\stlport\stlport
> Unquote. Which is it? The statements are contradictory. Are these two
> statements referring to two different iostreams, STLport's and some other
> one, or do they both refer to the STLport iostream? What is the "static
> library"?
>
> Following Christof's instructions I have:
> (built) the STLport libraries and dll's with: smake -f dm.mak
>
> bye, bye,
> Jim J.
>
> "Walter" <walter@digitalmars.com> wrote in message news:b4jotn$bfo$2@digitaldaemon.com...
> > Try the version of <iostream> in \dm\stlport\stlport, the older
> <iostream.h>
> > in \dm\include is obsolete (but included for backwards compabitility).
> >
> > "Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b4harj$2fmg$1@digitaldaemon.com...
> > > The in_file.bad() instruction in the code below does not return true
if
> a
> > > bad file name is entered on the command line. The cerr message and the return are skipped. The two commented lines return true, and the error message and the return statement execute for both of them. I am using
> the
> > > stlport library with dmc.
> > > According to N. Josuttis, C++ Standard Library, p.598, bad() is
supposed
> > to
> > > return true if  a fatal error has occurred (badbit is set). I have
> > compiled
> > > this program with dmc and another compiler with identical results.
> > >
> > >     std::ifstream in_file(argv[1]);
> > > //    if (in_file.fail())
> > > //    if (!in_file)
> > >     if (in_file.bad())
> > >     {
> > >          std::cerr << "Cannot open " << argv[1] << '\n';
> > >          return -1;
> > >     }
> > >
> > > I can find bad() in dm\include\iostream.h and some dll's in dm\bin
> > > iostream.h says this:
> > > inline int      ios::fail() { return state & (failbit | badbit |
> > > hardfail); }
> > > inline int      ios::bad() { return state & (badbit | hardfail); }
> > >
> > > Am I correct in concluding that trying to open a non-existing input
file
> > > does not turn on either the badbit or the hardfail, but only the
> failbit?
> > > Jim W. J.
>
>
>

I am still confused. I will read the stlport docs to see if I can make sense
of it.
I have gone back to the original sc.ini file, and I am
using -I\dm\stlport\stlport when compiling. The way I read the stlport
readme.txt file, that will allow me to link with the stlport static library,
and use the stlport iostreams.
No matter how I do it, the iostream bad() function does not work. But it
does not work for g++, or Borland either. You have to use the fail()
function to detect a file-open error.
I do not want to bother you with this anymore.

jwj.

"Walter" <walter@digitalmars.com> wrote in message news:b4l4im$2ae3$1@digitaldaemon.com...
> stlport can be used with and without iostreams. To use the iostreams,
you'll
> need to link with the stlport static library. If using stlport without iostreams, you don't need to link with that library. Sorry about the confusion. -Walter
>

"Jim Jennings" <jwjenn@mindspring.com> wrote in message news:b4ldg4$2huh$1@digitaldaemon.com...
> I am still confused. I will read the stlport docs to see if I can make
sense
> of it.
> I have gone back to the original sc.ini file, and I am
> using -I\dm\stlport\stlport when compiling. The way I read the stlport
> readme.txt file, that will allow me to link with the stlport static
library,
> and use the stlport iostreams.
> No matter how I do it, the iostream bad() function does not work. But it
> does not work for g++, or Borland either. You have to use the fail()
> function to detect a file-open error.
> I do not want to bother you with this anymore.

I am also trying to use stlport as generically as possible with as few changes to the source code in it as possible. I'm a bit reluctant to change the semantics of it as several compilers use it, and programmers tend to expect it to work the same on each.