Hello again!

I thought I should mention that I ran into a small issue with MinGW 3.1.0 when compiling a program that used the WinSTL registry functionality.  In the included version of the standard library, std::wstring doesn't look like it gets defined, but the compiler winds up hitting the #else half of the "#if defined" listed below.

#ifndef _STLSOFT_STRING_ACCESS_NO_STD_STRING
 #if defined(__STLSOFT_COMPILER_IS_GCC) && \
     __GNUC__ < 3
  typedef stlsoft_ns_qual_std(basic_string)<ss_char_w_t>
_stlsoft_wstring_t;
 #else
  typedef stlsoft_ns_qual_std(wstring)
_stlsoft_wstring_t;

Since std::wstring doesn't get defined for me, presumably because I'm not using STLPort, I got a long series of errors.

In my program, I worked around this by including the following in my source
code (this is by no means a permanent fix, but I stuck it in there to get me
off the ground):
#include <string>
namespace std
{
 typedef basic_string<wchar_t> wstring;
}

For the purposes of mothballing this problem, you can use reg_key_sequence_test.cpp.  Here's a version of reg_key_sequence_test.cpp that includes the quick fix I used before.

Anyway, this problem is easy to work around, but I wanted to stick a note on here before I forgot about it.

// This will cause various compile-time messages to be emitted. When you get
// sick of them just comment out or remove this #define
#define _STLSOFT_COMPILE_VERBOSE

#ifdef __DMC__
#define processheap_allocator                   _pha
#define reg_traits                              _rt
#define basic_reg_key                           _brk
#define basic_reg_key_sequence_const_iterator   _brksci
#endif /* __DMC__ */

#include <winstl.h>

#include <string>
namespace std
{
 typedef basic_string<wchar_t> wstring;
}

#include <winstl_reg_key_sequence.h>

#include <stdio.h>

#include <algorithm>
#include <functional>

winstl_ns_using(ws_char_a_t)
winstl_ns_using(reg_key_a)
winstl_ns_using(reg_key_sequence_a)


static void iterate_recurse(reg_key_sequence_a::const_iterator from, reg_key_sequence_a::const_iterator to, int depth);


struct dump_key
 : public winstl_ns_qual_std(unary_function)<reg_key_a const &, void>
{
 void operator ()(reg_key_a const key)
 {
  printf("%s\n", key.name().c_str());
 }
};


int main(int /* argc */, char ** /*argv*/)
{
 reg_key_sequence_a keys(HKEY_CURRENT_USER, "SOFTWARE\\Microsoft");

 printf("Iterating non-recursively (forward):\n");
 winstl_ns_qual_std(for_each)(keys.begin(), keys.end(), dump_key());

#if defined(__STLSOFT_CF_BIDIRECTIONAL_ITERATOR_SUPPORT)
    printf("Iterating non-recursively (backward):\n");
 winstl_ns_qual_std(for_each)(keys.rbegin(), keys.rend(), dump_key());
#endif /* __STLSOFT_CF_BIDIRECTIONAL_ITERATOR_SUPPORT */

 printf("Iterating recursively:\n");
 iterate_recurse(keys.begin(), keys.end(), 1);

    return 0;
}


void iterate_recurse(reg_key_sequence_a::const_iterator from,
reg_key_sequence_a::const_iterator to, int depth)
{
 ws_char_a_t prefix[255];

 memset(prefix, ' ', sizeof(prefix));
 prefix[depth] = '\0';

 for(; from != to; ++from)
 {
  const reg_key_a  &key = *from;
  reg_key_sequence_a keys(key);

  printf("%s, %s\n", prefix, key.name().c_str());

  iterate_recurse(keys.begin(), keys.end(), 1 + depth);
 }
}

/* ////////////////////////////////////////////////////////////////////////// */


#if 0 "basic_reg_key_sequence_const_iterator<>::basic_reg_key_sequence_const_itera tor<>(struct HKEY__ *)"

#endif /* 0 */

I can't download anything from MinGW other than 3.2.3 (although it's listed as MinGW-3.1.0-1.exe(, Is that the one you used?

If not, do you have the url you used to get the installer for your version?

It's going to be pretty messy trying to maintain compatibilities, unless I can get the one you've used.

"John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmt55a$d4p$1@digitaldaemon.com...
> Hello again!
>
> I thought I should mention that I ran into a small issue with MinGW 3.1.0 when compiling a program that used the WinSTL registry functionality.  In the included version of the standard library, std::wstring doesn't look
like
> it gets defined, but the compiler winds up hitting the #else half of the "#if defined" listed below.
>
> #ifndef _STLSOFT_STRING_ACCESS_NO_STD_STRING
>  #if defined(__STLSOFT_COMPILER_IS_GCC) && \
>      __GNUC__ < 3
>   typedef stlsoft_ns_qual_std(basic_string)<ss_char_w_t>
> _stlsoft_wstring_t;
>  #else
>   typedef stlsoft_ns_qual_std(wstring)
> _stlsoft_wstring_t;
>
> Since std::wstring doesn't get defined for me, presumably because I'm not using STLPort, I got a long series of errors.
>
> In my program, I worked around this by including the following in my
source
> code (this is by no means a permanent fix, but I stuck it in there to get
me
> off the ground):
> #include <string>
> namespace std
> {
>  typedef basic_string<wchar_t> wstring;
> }
>
> For the purposes of mothballing this problem, you can use reg_key_sequence_test.cpp.  Here's a version of reg_key_sequence_test.cpp that includes the quick fix I used before.
>
> Anyway, this problem is easy to work around, but I wanted to stick a note
on
> here before I forgot about it.
>
> // This will cause various compile-time messages to be emitted. When you
get
> // sick of them just comment out or remove this #define #define _STLSOFT_COMPILE_VERBOSE
>
> #ifdef __DMC__
> #define processheap_allocator                   _pha
> #define reg_traits                              _rt
> #define basic_reg_key                           _brk
> #define basic_reg_key_sequence_const_iterator   _brksci
> #endif /* __DMC__ */
>
> #include <winstl.h>
>
> #include <string>
> namespace std
> {
>  typedef basic_string<wchar_t> wstring;
> }
>
> #include <winstl_reg_key_sequence.h>
>
> #include <stdio.h>
>
> #include <algorithm>
> #include <functional>
>
> winstl_ns_using(ws_char_a_t)
> winstl_ns_using(reg_key_a)
> winstl_ns_using(reg_key_sequence_a)
>
>
> static void iterate_recurse(reg_key_sequence_a::const_iterator from, reg_key_sequence_a::const_iterator to, int depth);
>
>
> struct dump_key
>  : public winstl_ns_qual_std(unary_function)<reg_key_a const &, void>
> {
>  void operator ()(reg_key_a const key)
>  {
>   printf("%s\n", key.name().c_str());
>  }
> };
>
>
> int main(int /* argc */, char ** /*argv*/)
> {
>  reg_key_sequence_a keys(HKEY_CURRENT_USER, "SOFTWARE\\Microsoft");
>
>  printf("Iterating non-recursively (forward):\n");
>  winstl_ns_qual_std(for_each)(keys.begin(), keys.end(), dump_key());
>
> #if defined(__STLSOFT_CF_BIDIRECTIONAL_ITERATOR_SUPPORT)
>     printf("Iterating non-recursively (backward):\n");
>  winstl_ns_qual_std(for_each)(keys.rbegin(), keys.rend(), dump_key());
> #endif /* __STLSOFT_CF_BIDIRECTIONAL_ITERATOR_SUPPORT */
>
>  printf("Iterating recursively:\n");
>  iterate_recurse(keys.begin(), keys.end(), 1);
>
>     return 0;
> }
>
>
> void iterate_recurse(reg_key_sequence_a::const_iterator from,
> reg_key_sequence_a::const_iterator to, int depth)
> {
>  ws_char_a_t prefix[255];
>
>  memset(prefix, ' ', sizeof(prefix));
>  prefix[depth] = '\0';
>
>  for(; from != to; ++from)
>  {
>   const reg_key_a  &key = *from;
>   reg_key_sequence_a keys(key);
>
>   printf("%s, %s\n", prefix, key.name().c_str());
>
>   iterate_recurse(keys.begin(), keys.end(), 1 + depth);
>  }
> }
>
> /* ////////////////////////////////////////////////////////////////////////// */
>
>
> #if 0
>
"basic_reg_key_sequence_const_iterator<>::basic_reg_key_sequence_const_itera
> tor<>(struct HKEY__ *)"
>
> #endif /* 0 */
>
>

> I can't download anything from MinGW other than 3.2.3 (although it's
listed
> as MinGW-3.1.0-1.exe(, Is that the one you used?
That's the link that I used, and in my C:\MinGW\doc\MinGW, the docs all indicate that I'm running MinGW-3.1.0.  Is there some other way to check? Either way, that's the link that I used.  Maybe MinGW and gcc run different version schemes?

Let me know if there's anything else you need to know about my installation.

Also, I see that MinGW is not explicitly supported by STLSoft, so there's probably no point in you worrying about this.  I'd always thought that gcc and MinGW were very closely tied together, but it looks like things may not be packaged the same way.

g++ --version

Although maybe we're talking at cross purposes. It's possible that MinGW discuss a separate version to the version of GCC.

Anyway, if this is the version you have, I have enough to go on, so I'll give it a bash.

:)

"John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmuk70$27df$1@digitaldaemon.com...
> > I can't download anything from MinGW other than 3.2.3 (although it's
> listed
> > as MinGW-3.1.0-1.exe(, Is that the one you used?
> That's the link that I used, and in my C:\MinGW\doc\MinGW, the docs all indicate that I'm running MinGW-3.1.0.  Is there some other way to check? Either way, that's the link that I used.  Maybe MinGW and gcc run
different
> version schemes?
>
> Let me know if there's anything else you need to know about my
installation.
>
> Also, I see that MinGW is not explicitly supported by STLSoft, so there's probably no point in you worrying about this.  I'd always thought that gcc and MinGW were very closely tied together, but it looks like things may
not
> be packaged the same way.
>
>
>

Yep, you've got it right - my g++ is 3.2.3, and MinGW docs claim to be 3.1.0.  Apparently, it looks like MinGW might package the standard library differently than plain ol' g++ - which would be quite painful.

Anyway, I realize MinGW's not officially supported, so don't bend over backwards on this one.  I can easily work around this problem and note anything else that I might run into.

Thanks!

"Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmuq96$2f0g$1@digitaldaemon.com...
> g++ --version
>
> Although maybe we're talking at cross purposes. It's possible that MinGW discuss a separate version to the version of GCC.
>
> Anyway, if this is the version you have, I have enough to go on, so I'll give it a bash.
>
> :)
>
> "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmuk70$27df$1@digitaldaemon.com...
> > > I can't download anything from MinGW other than 3.2.3 (although it's
> > listed
> > > as MinGW-3.1.0-1.exe(, Is that the one you used?
> > That's the link that I used, and in my C:\MinGW\doc\MinGW, the docs all indicate that I'm running MinGW-3.1.0.  Is there some other way to
check?
> > Either way, that's the link that I used.  Maybe MinGW and gcc run
> different
> > version schemes?
> >
> > Let me know if there's anything else you need to know about my
> installation.
> >
> > Also, I see that MinGW is not explicitly supported by STLSoft, so
there's
> > probably no point in you worrying about this.  I'd always thought that
gcc
> > and MinGW were very closely tied together, but it looks like things may
> not
> > be packaged the same way.
> >
> >
> >
>
>

Well, MinGW is my prime source of GCC, since I'm wont to boot the Linux box only when I absolutely *have* to. So in a sense it's fair to say that STLSoft does support GCC. However, if you look at the compiler compatibility tables on the site, it should just say GCC 2.95, 2.96 and 3.2. So I could weasel out by saying 3.2.3 is not supported. However,  the only reason stopping me sorting this compatibility is time, as I have 10 weeks to complete the book. Hence, your may be working around with your workaround for some time ...

Don't hesitate to remind me about this in a few weeks. ;)

"John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmut6u$2ire$1@digitaldaemon.com...
> Yep, you've got it right - my g++ is 3.2.3, and MinGW docs claim to be 3.1.0.  Apparently, it looks like MinGW might package the standard library differently than plain ol' g++ - which would be quite painful.
>
> Anyway, I realize MinGW's not officially supported, so don't bend over backwards on this one.  I can easily work around this problem and note anything else that I might run into.
>
> Thanks!
>
> "Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmuq96$2f0g$1@digitaldaemon.com...
> > g++ --version
> >
> > Although maybe we're talking at cross purposes. It's possible that MinGW discuss a separate version to the version of GCC.
> >
> > Anyway, if this is the version you have, I have enough to go on, so I'll give it a bash.
> >
> > :)
> >
> > "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmuk70$27df$1@digitaldaemon.com...
> > > > I can't download anything from MinGW other than 3.2.3 (although it's
> > > listed
> > > > as MinGW-3.1.0-1.exe(, Is that the one you used?
> > > That's the link that I used, and in my C:\MinGW\doc\MinGW, the docs
all
> > > indicate that I'm running MinGW-3.1.0.  Is there some other way to
> check?
> > > Either way, that's the link that I used.  Maybe MinGW and gcc run
> > different
> > > version schemes?
> > >
> > > Let me know if there's anything else you need to know about my
> > installation.
> > >
> > > Also, I see that MinGW is not explicitly supported by STLSoft, so
> there's
> > > probably no point in you worrying about this.  I'd always thought that
> gcc
> > > and MinGW were very closely tied together, but it looks like things
may
> > not
> > > be packaged the same way.
> > >
> > >
> > >
> >
> >
>
>

My assessment - write the book!  This is no big deal at all, and I'm sure that going with STLPort would fix it in an instant.  My workaround works just fine for now, and that book isn't going to write itself.  :)

Good luck with the writing!

"Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmv03n$2mhd$1@digitaldaemon.com...
> Well, MinGW is my prime source of GCC, since I'm wont to boot the Linux
box
> only when I absolutely *have* to. So in a sense it's fair to say that STLSoft does support GCC. However, if you look at the compiler
compatibility
> tables on the site, it should just say GCC 2.95, 2.96 and 3.2. So I could weasel out by saying 3.2.3 is not supported. However,  the only reason stopping me sorting this compatibility is time, as I have 10 weeks to complete the book. Hence, your may be working around with your workaround for some time ...
>
> Don't hesitate to remind me about this in a few weeks. ;)
>
> "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmut6u$2ire$1@digitaldaemon.com...
> > Yep, you've got it right - my g++ is 3.2.3, and MinGW docs claim to be 3.1.0.  Apparently, it looks like MinGW might package the standard
library
> > differently than plain ol' g++ - which would be quite painful.
> >
> > Anyway, I realize MinGW's not officially supported, so don't bend over backwards on this one.  I can easily work around this problem and note anything else that I might run into.
> >
> > Thanks!
> >
> > "Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmuq96$2f0g$1@digitaldaemon.com...
> > > g++ --version
> > >
> > > Although maybe we're talking at cross purposes. It's possible that
MinGW
> > > discuss a separate version to the version of GCC.
> > >
> > > Anyway, if this is the version you have, I have enough to go on, so
I'll
> > > give it a bash.
> > >
> > > :)
> > >
> > > "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmuk70$27df$1@digitaldaemon.com...
> > > > > I can't download anything from MinGW other than 3.2.3 (although
it's
> > > > listed
> > > > > as MinGW-3.1.0-1.exe(, Is that the one you used?
> > > > That's the link that I used, and in my C:\MinGW\doc\MinGW, the docs
> all
> > > > indicate that I'm running MinGW-3.1.0.  Is there some other way to
> > check?
> > > > Either way, that's the link that I used.  Maybe MinGW and gcc run
> > > different
> > > > version schemes?
> > > >
> > > > Let me know if there's anything else you need to know about my
> > > installation.
> > > >
> > > > Also, I see that MinGW is not explicitly supported by STLSoft, so
> > there's
> > > > probably no point in you worrying about this.  I'd always thought
that
> > gcc
> > > > and MinGW were very closely tied together, but it looks like things
> may
> > > not
> > > > be packaged the same way.
> > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>

Cheers mate. ;)


"John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmv1e0$2o56$1@digitaldaemon.com...
> My assessment - write the book!  This is no big deal at all, and I'm sure that going with STLPort would fix it in an instant.  My workaround works just fine for now, and that book isn't going to write itself.  :)
>
> Good luck with the writing!
>
> "Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmv03n$2mhd$1@digitaldaemon.com...
> > Well, MinGW is my prime source of GCC, since I'm wont to boot the Linux
> box
> > only when I absolutely *have* to. So in a sense it's fair to say that STLSoft does support GCC. However, if you look at the compiler
> compatibility
> > tables on the site, it should just say GCC 2.95, 2.96 and 3.2. So I
could
> > weasel out by saying 3.2.3 is not supported. However,  the only reason stopping me sorting this compatibility is time, as I have 10 weeks to complete the book. Hence, your may be working around with your
workaround
> > for some time ...
> >
> > Don't hesitate to remind me about this in a few weeks. ;)
> >
> > "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmut6u$2ire$1@digitaldaemon.com...
> > > Yep, you've got it right - my g++ is 3.2.3, and MinGW docs claim to be 3.1.0.  Apparently, it looks like MinGW might package the standard
> library
> > > differently than plain ol' g++ - which would be quite painful.
> > >
> > > Anyway, I realize MinGW's not officially supported, so don't bend over backwards on this one.  I can easily work around this problem and note anything else that I might run into.
> > >
> > > Thanks!
> > >
> > > "Matthew Wilson" <matthew@stlsoft.org> wrote in message news:bmuq96$2f0g$1@digitaldaemon.com...
> > > > g++ --version
> > > >
> > > > Although maybe we're talking at cross purposes. It's possible that
> MinGW
> > > > discuss a separate version to the version of GCC.
> > > >
> > > > Anyway, if this is the version you have, I have enough to go on, so
> I'll
> > > > give it a bash.
> > > >
> > > > :)
> > > >
> > > > "John Abney" <johnNOSPAMPLEASEabney@yahoo.com> wrote in message news:bmuk70$27df$1@digitaldaemon.com...
> > > > > > I can't download anything from MinGW other than 3.2.3 (although
> it's
> > > > > listed
> > > > > > as MinGW-3.1.0-1.exe(, Is that the one you used?
> > > > > That's the link that I used, and in my C:\MinGW\doc\MinGW, the
docs
> > all
> > > > > indicate that I'm running MinGW-3.1.0.  Is there some other way to
> > > check?
> > > > > Either way, that's the link that I used.  Maybe MinGW and gcc run
> > > > different
> > > > > version schemes?
> > > > >
> > > > > Let me know if there's anything else you need to know about my
> > > > installation.
> > > > >
> > > > > Also, I see that MinGW is not explicitly supported by STLSoft, so
> > > there's
> > > > > probably no point in you worrying about this.  I'd always thought
> that
> > > gcc
> > > > > and MinGW were very closely tied together, but it looks like
things
> > may
> > > > not
> > > > > be packaged the same way.
> > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
>