I've been looking through Tango documentation for a function that'll tell me the total amount of RAM installed on the system, bu I can't find anything that looks helpful. Anyone have any suggestions?

Thanks!

--benji

"Benji Smith" wrote
> I've been looking through Tango documentation for a function that'll tell me the total amount of RAM installed on the system, bu I can't find anything that looks helpful. Anyone have any suggestions?
>
> Thanks!

This is something that is highly OS-dependant (and even different versions of the same OS do it differently).  It's not something that Tango is likely to provide an abstract interface for.

-Steve

Steven Schveighoffer wrote:
> "Benji Smith" wrote
>> I've been looking through Tango documentation for a function that'll tell me the total amount of RAM installed on the system, bu I can't find anything that looks helpful. Anyone have any suggestions?
>>
>> Thanks!
> 
> This is something that is highly OS-dependant (and even different versions of the same OS do it differently).  It's not something that Tango is likely to provide an abstract interface for.
> 
> -Steve 

Okay. I suppose I'll have to find the win32 & linux functions and write the wrappers myself.

Thanks!

--benji

On Wed, 13 Aug 2008 22:39:36 +0400, Steven Schveighoffer <schveiguy@yahoo.com> wrote:

> "Benji Smith" wrote
>> I've been looking through Tango documentation for a function that'll tell
>> me the total amount of RAM installed on the system, bu I can't find
>> anything that looks helpful. Anyone have any suggestions?
>>
>> Thanks!
>
> This is something that is highly OS-dependant (and even different versions
> of the same OS do it differently).  It's not something that Tango is likely
> to provide an abstract interface for.
>
> -Steve
>
>

The same goes for IO (console I/O, file I/O, socket etc). Despite that it is present in Tango.

I see a huge benefit of having (at least basic) OS statistics (like OS name, version, type, device presence and capabilities - RAM, CPU, etc) in the standard library. Isn't is supposed to eliminate custom platform-specific code?

"Denis Koroskin" wrote
> On Wed, 13 Aug 2008 22:39:36 +0400, Steven Schveighoffer
>> "Benji Smith" wrote
>>> I've been looking through Tango documentation for a function that'll
>>> tell
>>> me the total amount of RAM installed on the system, bu I can't find
>>> anything that looks helpful. Anyone have any suggestions?
>>>
>>> Thanks!
>>
>> This is something that is highly OS-dependant (and even different
>> versions
>> of the same OS do it differently).  It's not something that Tango is
>> likely
>> to provide an abstract interface for.
>>
>> -Steve
>>
>>
>
> The same goes for IO (console I/O, file I/O, socket etc). Despite that it is present in Tango.
>
> I see a huge benefit of having (at least basic) OS statistics (like OS name, version, type, device presence and capabilities - RAM, CPU, etc) in the standard library. Isn't is supposed to eliminate custom platform-specific code?

Getting system memory isn't part of any standard that I know of (at least on the posix side), and there's not a large demand for it.  The API for getting memory is decided on by OS implementors.  What you would end up with is a list of versions for not only different OSes, but different versions of the OS.  It's along the same lines as getting/setting network interface addresses.

From experience (I have written a program to get system memory size), it is not trivial.  And not always accurate.

This is not to say that it couldn't be added.  If you come up with something that looks reasonable, Tango might accept it.

-Steve

Steven Schveighoffer wrote:
> Getting system memory isn't part of any standard that I know of (at least on the posix side), and there's not a large demand for it.  The API for getting memory is decided on by OS implementors.  What you would end up with is a list of versions for not only different OSes, but different versions of the OS.  It's along the same lines as getting/setting network interface addresses.
> 
> From experience (I have written a program to get system memory size), it is not trivial.  And not always accurate.
> 
> This is not to say that it couldn't be added.  If you come up with something that looks reasonable, Tango might accept it.
> 
> -Steve 


In win32, it looks like the GlobalMemoryStatus (for all non-NT versions) and GlobalMemoryStatusEx (for all NT-based versions), from kernel32.dll, would do the trick.

http://msdn.microsoft.com/en-us/library/aa366586.aspx
http://msdn.microsoft.com/en-us/library/aa366589(VS.85).aspx

In linux, a bare-minimum solution would be to read from /proc/meminfo, even if there's no POSIX function.

Of course, I'm talking out of my ass here, because my experience with C/C++ programming is very thin. I don't know, for example, how you'd check for the existence of the GlobalMemoryStatusEx function and fall back to GlobalMemoryStatus when it doesn't exist.

It seems doable, though my own systems programming experience is probably too lacking for me to make it happen myself.

--benji

Benji Smith wrote:
> Steven Schveighoffer wrote:
>> Getting system memory isn't part of any standard that I know of (at least on the posix side), and there's not a large demand for it.  The API for getting memory is decided on by OS implementors.  What you would end up with is a list of versions for not only different OSes, but different versions of the OS.  It's along the same lines as getting/setting network interface addresses.
>>
>> From experience (I have written a program to get system memory size), it is not trivial.  And not always accurate.
>>
>> This is not to say that it couldn't be added.  If you come up with something that looks reasonable, Tango might accept it.
>>
>> -Steve 
> 
> 
> In win32, it looks like the GlobalMemoryStatus (for all non-NT versions) and GlobalMemoryStatusEx (for all NT-based versions), from kernel32.dll, would do the trick.
> 
> http://msdn.microsoft.com/en-us/library/aa366586.aspx
> http://msdn.microsoft.com/en-us/library/aa366589(VS.85).aspx
> 
> In linux, a bare-minimum solution would be to read from /proc/meminfo, even if there's no POSIX function.
> 
> Of course, I'm talking out of my ass here, because my experience with C/C++ programming is very thin. I don't know, for example, how you'd check for the existence of the GlobalMemoryStatusEx function and fall back to GlobalMemoryStatus when it doesn't exist.
> 
> It seems doable, though my own systems programming experience is probably too lacking for me to make it happen myself.
> 
> --benji

Id default to NT version since its Win 2000+,
Then force version flag for non-NT,  or dont support 95, 98, ME

version(WIN_NO_NT)
  alias GlobalMemoryStatus GlobalMemoryStat;
else
  alias GlobalMemoryStatusEx GlobalMemoryStat;

Wyverex wrote:
> Id default to NT version since its Win 2000+,
> Then force version flag for non-NT,  or dont support 95, 98, ME
> 
> version(WIN_NO_NT)
>   alias GlobalMemoryStatus GlobalMemoryStat;
> else
>   alias GlobalMemoryStatusEx GlobalMemoryStat;

For this project, I'm trying to support all versions of windows, even those old beasties 95, 98, and ME.

(The project is a library for reporting anonymous statistics about user behavior. Failing to support the older operating systems would create a bias in the reporting that I'd rather avoid.)

And, if at all possible, I'd rather just compile a single binary, even if it means jumping through some weird hoops in writing my code. Is there any feasible solution?

--benji

Benji Smith wrote:
> Wyverex wrote:
>> Id default to NT version since its Win 2000+,
>> Then force version flag for non-NT,  or dont support 95, 98, ME
>>
>> version(WIN_NO_NT)
>>   alias GlobalMemoryStatus GlobalMemoryStat;
>> else
>>   alias GlobalMemoryStatusEx GlobalMemoryStat;
> 
> For this project, I'm trying to support all versions of windows, even those old beasties 95, 98, and ME.
> 
> (The project is a library for reporting anonymous statistics about user behavior. Failing to support the older operating systems would create a bias in the reporting that I'd rather avoid.)
> 
> And, if at all possible, I'd rather just compile a single binary, even if it means jumping through some weird hoops in writing my code. Is there any feasible solution?
> 
> --benji

Looks like you'd have to try to get the function address from the DLL..

http://msdn.microsoft.com/en-us/library/ms724832(VS.85).aspx

"Benji Smith" wrote
> Wyverex wrote:
>> Id default to NT version since its Win 2000+,
>> Then force version flag for non-NT,  or dont support 95, 98, ME
>>
>> version(WIN_NO_NT)
>>   alias GlobalMemoryStatus GlobalMemoryStat;
>> else
>>   alias GlobalMemoryStatusEx GlobalMemoryStat;
>
> For this project, I'm trying to support all versions of windows, even those old beasties 95, 98, and ME.
>
> (The project is a library for reporting anonymous statistics about user behavior. Failing to support the older operating systems would create a bias in the reporting that I'd rather avoid.)
>
> And, if at all possible, I'd rather just compile a single binary, even if it means jumping through some weird hoops in writing my code. Is there any feasible solution?

Usually, the non-EX versions of MS functions work fine on the newer platforms, they are just not preferred.

Unless, of course, there is some resolution issue :)

In that case, you can call the GetVersion function. http://msdn.microsoft.com/en-us/library/ms724439.aspx

-Steve