Digital Mars - c++ - 8.1e - mutable doesn't work!

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

The "mutable" keyword apparently doesn't work...

class Test
{
  public:
    void modify() const { c = 1.0; }
  private:
    mutable double c;
};

I'm getting the error "cannot modify const variable"... 
Unfortunaltely, I really need mutable to work in my application -
Walter, what do you mean by "mutable implemented" in 8.1e ChangeLog?


Laurentiu

Sep 28 2001

↑ ↓ ← → Christof Meerwald <NOSPAM_seeMySig fastrun.at> writes:

On Fri, 28 Sep 2001 17:50:50 GMT, Laurentiu Pancescu wrote:
 The "mutable" keyword apparently doesn't work...
 
 class Test
 {
   public:
     void modify() const { c = 1.0; }
   private:
     mutable double c;
 };


Hmm, works for me. Are you sure you are using 8.1e?


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw jabber.at
mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

Sep 28 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

Christof Meerwald <NOSPAM_seeMySig fastrun.at> wrote:

On Fri, 28 Sep 2001 17:50:50 GMT, Laurentiu Pancescu wrote:
 The "mutable" keyword apparently doesn't work...
 
 class Test
 {
   public:
     void modify() const { c = 1.0; }
   private:
     mutable double c;
 };


Hmm, works for me. Are you sure you are using 8.1e?


Yep, it's 8.1e - I tried to reduce some more complicated code
from my application, but didn't try it...  It doesn't work in my
app, but in simple code yes!! :(

I'm using <iostream>, <iomanip> and STLport 4.0, but mutable
doesn't seem to be defined in there - I don't get what's going
on...  I'll try to see if I can reproduce it somehow.

Laurentiu

Sep 28 2001

↑ ↓ ← → Christof Meerwald <cmeerw web.de> writes:

On Fri, 28 Sep 2001 18:08:05 GMT, Laurentiu Pancescu wrote:
 I'm using <iostream>, <iomanip> and STLport 4.0, but mutable
 doesn't seem to be defined in there - I don't get what's going


There is a "#define mutable" in stlport/stl/_config.h which is used if
__STL_NEED_MUTABLE is defined in the compiler configuration file (just make
sure you don't define it there).

You could also try to generate a listing of the preprocessed source-code
using "-e -lfile.lst" and check if the mutable keyword is still there.


bye, Christof

-- 
http://cmeerw.cjb.net                          Jabber: cmeerw jabber.at
mailto cmeerw at web.de                   ICQ: 93773535, Yahoo!: cmeerw

...and what have you contributed to the Net?

Sep 28 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

Christof Meerwald <cmeerw web.de> wrote:

On Fri, 28 Sep 2001 18:08:05 GMT, Laurentiu Pancescu wrote:
 I'm using <iostream>, <iomanip> and STLport 4.0, but mutable
 doesn't seem to be defined in there - I don't get what's going


There is a "#define mutable" in stlport/stl/_config.h which is used if
__STL_NEED_MUTABLE is defined in the compiler configuration file (just make
sure you don't define it there).


I'm getting:

Internal error: cg87 263

Regards,
  Laurentiu

Sep 28 2001

↑ ↓ ← → "Walter" <walter digitalmars.com> writes:

Laurentiu Pancescu wrote in message <9p2fc2$26em$1 digitaldaemon.com>...
I'm getting:

Internal error: cg87 263



Any chance of getting a small, reproducible test case of that?

Sep 28 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

"Walter" <walter digitalmars.com> wrote:

Laurentiu Pancescu wrote in message <9p2fc2$26em$1 digitaldaemon.com>...
I'm getting:

Internal error: cg87 263



Any chance of getting a small, reproducible test case of that?


I'm afraid not... What's strange is that I added a line at the
very end of my file, containing '#error READY', and it works:
I'm getting the READY error, so I think that this internal error
doesn't occur in my code, but after finishing compiling the
file (huh?).

I wasn't able to reproduce this behavior with any small test
example, but, if you want, I can send you
(walter digitalmars.com, right?) a ZIP with a part of my program, enough to
produce
that error (a single cpp file, with some headers).  I'm using the
latest release of STLport, downloaded from digitalmars.com
(you need that for compiling).

Regards,
  Laurentiu

Sep 29 2001

↑ ↓ ← → "Walter" <walter digitalmars.com> writes:

All right. -Walter

Laurentiu Pancescu wrote in message <9p3uhd$30ai$1 digitaldaemon.com>...
"Walter" <walter digitalmars.com> wrote:

Laurentiu Pancescu wrote in message <9p2fc2$26em$1 digitaldaemon.com>...
I'm getting:

Internal error: cg87 263



Any chance of getting a small, reproducible test case of that?


I'm afraid not... What's strange is that I added a line at the
very end of my file, containing '#error READY', and it works:
I'm getting the READY error, so I think that this internal error
doesn't occur in my code, but after finishing compiling the
file (huh?).

I wasn't able to reproduce this behavior with any small test
example, but, if you want, I can send you
(walter digitalmars.com, right?) a ZIP with a part of my program, enough to


that error (a single cpp file, with some headers).  I'm using the
latest release of STLport, downloaded from digitalmars.com
(you need that for compiling).

Regards,
  Laurentiu

Sep 29 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

"Walter" <walter digitalmars.com> wrote:

All right. -Walter

Laurentiu Pancescu wrote in message <9p3uhd$30ai$1 digitaldaemon.com>...
"Walter" <walter digitalmars.com> wrote:

Laurentiu Pancescu wrote in message <9p2fc2$26em$1 digitaldaemon.com>...
I'm getting:

Internal error: cg87 263



Any chance of getting a small, reproducible test case of that?






I hope you can fix this soon... I'm writing a numerical
application for a friend (PhD student), who has a lot of calculations
to do.  Currently, it takes more than a week for a run (she
wrote it in TurboPascal ;), so any speed improvement is more than
welcome.  I'm currently using DJGPP to generate a DOS extended
app, since it's the fastest, but still DMC was about 50% faster
than this.  It could be about a few saved days at every run...

Thanks,
  Laurentiu

Sep 29 2001

↑ ↓ ← → "Walter" <walter digitalmars.com> writes:

Laurentiu Pancescu wrote in message <9p52um$l0p$1 digitaldaemon.com>...
I hope you can fix this soon... I'm writing a numerical
application for a friend (PhD student), who has a lot of calculations
to do.  Currently, it takes more than a week for a run (she
wrote it in TurboPascal ;), so any speed improvement is more than
welcome.  I'm currently using DJGPP to generate a DOS extended
app, since it's the fastest, but still DMC was about 50% faster
than this.  It could be about a few saved days at every run...



I can get it fixed faster if you can shrink it down ...

Sep 29 2001

↑ ↓ ← → Damian <damiandixon netscape.net> writes:

Hi,

You really need to be looking in the directory

   stlport/config

at the file 'stl_digital_mars.h' for the compiler configuration.

I have just got back from doing some consultancy work in Sweden and will 
be looking at updating STLport 4.0 for the new compiler. I will then be 
porting STLport 4.5...

As such I will see if I can narrow the mutable problem down in the 
STLport library. Though I can't promise any timescales.

Regards
Damian


Christof Meerwald wrote:

 On Fri, 28 Sep 2001 18:08:05 GMT, Laurentiu Pancescu wrote:
 
I'm using <iostream>, <iomanip> and STLport 4.0, but mutable
doesn't seem to be defined in there - I don't get what's going


 There is a "#define mutable" in stlport/stl/_config.h which is used if
 __STL_NEED_MUTABLE is defined in the compiler configuration file (just make
 sure you don't define it there).
 
 You could also try to generate a listing of the preprocessed source-code
 using "-e -lfile.lst" and check if the mutable keyword is still there.
 
 
 bye, Christof

Oct 01 2001

↑ ↓ ← → "Walter" <walter digitalmars.com> writes:

Thanks! My current focus is getting the regular STL 3.3 to work, but I'll
see if any problems you isolate down can be fixed. -Walter

Damian wrote in message <3BB890D5.2080509 netscape.net>...
Hi,

You really need to be looking in the directory

   stlport/config

at the file 'stl_digital_mars.h' for the compiler configuration.

I have just got back from doing some consultancy work in Sweden and will
be looking at updating STLport 4.0 for the new compiler. I will then be
porting STLport 4.5...

As such I will see if I can narrow the mutable problem down in the
STLport library. Though I can't promise any timescales.

Regards
Damian


Christof Meerwald wrote:

 On Fri, 28 Sep 2001 18:08:05 GMT, Laurentiu Pancescu wrote:

I'm using <iostream>, <iomanip> and STLport 4.0, but mutable
doesn't seem to be defined in there - I don't get what's going


 There is a "#define mutable" in stlport/stl/_config.h which is used if
 __STL_NEED_MUTABLE is defined in the compiler configuration file (just




 sure you don't define it there).

 You could also try to generate a listing of the preprocessed source-code
 using "-e -lfile.lst" and check if the mutable keyword is still there.


 bye, Christof

Oct 01 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

Damian <damiandixon netscape.net> wrote:

Hi,

You really need to be looking in the directory

   stlport/config

at the file 'stl_digital_mars.h' for the compiler configuration.

I have just got back from doing some consultancy work in Sweden and will 
be looking at updating STLport 4.0 for the new compiler. I will then be 
porting STLport 4.5...

As such I will see if I can narrow the mutable problem down in the 
STLport library. Though I can't promise any timescales.


Hi, and thanks!

There's no mutable problem in STLport, or none that I know
of...  Everything works fine wrt STL, I initially thought mutable
generates that internal error.  But it seems to be something not
related to STL, but to floating point code generation (that's
what cg87 name suggests, maybe I'm wrong).

Thanks again,
  Laurentiu

Oct 01 2001

↑ ↓ ← → Laurentiu Pancescu <plaur crosswinds.net> writes:

Christof Meerwald <NOSPAM_seeMySig fastrun.at> wrote:

On Fri, 28 Sep 2001 17:50:50 GMT, Laurentiu Pancescu wrote:
 The "mutable" keyword apparently doesn't work...
 
 class Test
 {
   public:
     void modify() const { c = 1.0; }
   private:
     mutable double c;
 };


Hmm, works for me. Are you sure you are using 8.1e?


Okay, it seems that STLport is #defining mutable to nothing...
try including <vector> at the top of the file.  Probably
STLport modifications for DMC need some cleanup.  A simple '#undef
mutable' solves the error, but I don't want to clutter my code
with unnecessay #undefs, eventually depending on the compiler...


Laurentiu

Sep 28 2001