typedef for portability

Gregry2

I'm starting work on my bigint thing again...and I'm shooting for portability. Now, the question is on the types. Now I know that redefining types for the sake of portability is important for this...

However, I have a problem that might popup...this bigint things is obviously a component, so, since its compiled with other components, or its host program, it might run into a typdef of the same name. Such is of greater possibility if I use common names, like uint16 or so...

Any solution or idea, or should I just make a more safe name (like sint_t_unit16 or something...)

{2}rIng

MT2002

: I'm starting work on my bigint thing again...and I'm shooting for
: portability. Now, the question is on the types. Now I know that
: redefining types for the sake of portability is important for this...
:
: However, I have a problem that might popup...this bigint things is
: obviously a component, so, since its compiled with other components,
: or its host program, it might run into a typdef of the same name.
: Such is of greater possibility if I use common names, like uint16 or
: so...
:
: Any solution or idea, or should I just make a more safe name (like
: sint_t_unit16 or something...)
:
: {2}rIng

If you come accross a component are something that might cause naming
conflicts, just use the preprocessor to hide your typedef names,
or #undef them before you typedef them.

I personally append a _t to typedef type names, but then again
uint16_t is just as common as uint16.

Gregry2

: : I'm starting work on my bigint thing again...and I'm shooting for
: : portability. Now, the question is on the types. Now I know that
: : redefining types for the sake of portability is important for this...
: :
: : However, I have a problem that might popup...this bigint things is
: : obviously a component, so, since its compiled with other components,
: : or its host program, it might run into a typdef of the same name.
: : Such is of greater possibility if I use common names, like uint16 or
: : so...
: :
: : Any solution or idea, or should I just make a more safe name (like
: : sint_t_unit16 or something...)
: :
: : {2}rIng
:
: If you come accross a component are something that might cause naming
: conflicts, just use the preprocessor to hide your typedef names,
: or #undef them before you typedef them.
:
: I personally append a _t to typedef type names, but then again
: uint16_t is just as common as uint16.

can you #undef typedef names...I thought they are compiler handled, but #undef is for preproccessor...

Anyhow, heres another predicament...See, lets say I want to return a int that acts as a bitset from a few functions...now, to do that, a certain number of bits is required for the return type...but the thing is the type is returned...so the user of te component must have this type defined for itself...

I tried solving it the stupid way, redesigning my functions so they don't have to return bitsets(making them only have one status to return, thus, a single value)...

{2}rIng

MT2002

: : : I'm starting work on my bigint thing again...and I'm shooting for
: : : portability. Now, the question is on the types. Now I know that
: : : redefining types for the sake of portability is important for this...
: : :
: : : However, I have a problem that might popup...this bigint things is
: : : obviously a component, so, since its compiled with other components,
: : : or its host program, it might run into a typdef of the same name.
: : : Such is of greater possibility if I use common names, like uint16 or
: : : so...
: : :
: : : Any solution or idea, or should I just make a more safe name (like
: : : sint_t_unit16 or something...)
: : :
: : : {2}rIng
: :
: : If you come accross a component are something that might cause naming
: : conflicts, just use the preprocessor to hide your typedef names,
: : or #undef them before you typedef them.
: :
: : I personally append a _t to typedef type names, but then again
: : uint16_t is just as common as uint16.
:
: can you #undef typedef names...I thought they are compiler handled,
: but #undef is for preproccessor...

[color=Blue]Im assuming you never #undef new before?[/color] ;-)

: Anyhow, heres another predicament...See, lets say I want to return a
: int that acts as a bitset from a few functions...now, to do that, a
: certain number of bits is required for the return type...but the
: thing is the type is returned...so the user of te component must
: have this type defined for itself...

[color=Blue]If a routine needs to return a bit pattern, decide how many bits
it should return. ...Lets say, 32.

Pick a data type thats 32bits (Integers on 32bt systems), and have it
return that type. Then mask the value to test the bits.

Im not sure what you mean by [b]the user of te component must
have this type defined for itself[/b] -- Why?
[/color]
: I tried solving it the stupid way, redesigning my functions so they
: don't have to return bitsets(making them only have one status to
: return, thus, a single value)...

[color=Blue]I coinsider that a better method. If you can find a method that
isnt platform specific, and easier to modify, use it![/color]

Gregry2

: : : : I'm starting work on my bigint thing again...and I'm shooting for
: : : : portability. Now, the question is on the types. Now I know that
: : : : redefining types for the sake of portability is important for this...
: : : :
: : : : However, I have a problem that might popup...this bigint things is
: : : : obviously a component, so, since its compiled with other components,
: : : : or its host program, it might run into a typdef of the same name.
: : : : Such is of greater possibility if I use common names, like uint16 or
: : : : so...
: : : :
: : : : Any solution or idea, or should I just make a more safe name (like
: : : : sint_t_unit16 or something...)
: : : :
: : : : {2}rIng
: : :
: : : If you come accross a component are something that might cause naming
: : : conflicts, just use the preprocessor to hide your typedef names,
: : : or #undef them before you typedef them.
: : :
: : : I personally append a _t to typedef type names, but then again
: : : uint16_t is just as common as uint16.
: :
: : can you #undef typedef names...I thought they are compiler handled,
: : but #undef is for preproccessor...
:
: [color=Blue]Im assuming you never #undef new before?[/color] ;-)

[color=Red]Or may be I don't really understand...can you pls explain what you meant by "#undef them before you typedef them"...I know how to hide..but...ugh sorry...[/color]


:
: : Anyhow, heres another predicament...See, lets say I want to return a
: : int that acts as a bitset from a few functions...now, to do that, a
: : certain number of bits is required for the return type...but the
: : thing is the type is returned...so the user of te component must
: : have this type defined for itself...
:
: [color=Blue]If a routine needs to return a bit pattern, decide how
: many bits
: it should return. ...Lets say, 32.
:
: Pick a data type thats 32bits (Integers on 32bt systems), and have it
: return that type. Then mask the value to test the bits.
:
: Im not sure what you mean by [b]the user of te component must
: have this type defined for itself[/b] -- Why?
: [/color]

[color=Red](the new tag thingy is..intresting)Screw what I said...it doesn't make sense what I was thinking(lol XD)...[/color]

: : I tried solving it the stupid way, redesigning my functions so they
: : don't have to return bitsets(making them only have one status to
: : return, thus, a single value)...
:
: [color=Blue]I coinsider that a better method. If you can find a
: method that
: isnt platform specific, and easier to modify, use it![/color]

I thought that this was messy, to redesign everything on the sake of one problem...well I am a noob...or idk.

Hows this

[code]
#if sizeof(unsigned short) >= 2
typedef _sint_uint16_ unsigned short

#elif sizeof(unsigned int) >= 2
typedef _sint_uint16_ unsigned int

#elif sizeof(unsigned long) >= 2
typedef _sint_uint16_ unsigned long

#else
#error "No primitive type has a byte length of at least 2. This component cannot complie on this platform."
#endif
[/code]

Is this hunky-dori?

thnx
{2}rIng

MT2002

: [code]:
: #if sizeof(unsigned short) >= 2
: typedef _sint_uint16_ unsigned short
:
: #elif sizeof(unsigned int) >= 2
: typedef _sint_uint16_ unsigned int
:
: #elif sizeof(unsigned long) >= 2
: typedef _sint_uint16_ unsigned long
:
: #else
: #error "No primitive type has a byte length of at least 2. This component cannot complie on this platform."
: #endif
: [/code]:
:
: Is this hunky-dori?
:
: thnx
: {2}rIng
[code][/code]
I never thought about that method :-)
[code][/code]
I useually do it based off platform.
[code][/code]
Or, just use the ANSI [b]stdint.h[/b] :-)
[code][/code]
As for undefining a symbol:
[code]
#undef foo
typedef unsigned int foo;
[/code]

ige

: : [code]: :
: : #if sizeof(unsigned short) >= 2
: : typedef _sint_uint16_ unsigned short
: :
: : #elif sizeof(unsigned int) >= 2
: : typedef _sint_uint16_ unsigned int
: :
: : #elif sizeof(unsigned long) >= 2
: : typedef _sint_uint16_ unsigned long
: :
: : #else
: : #error "No primitive type has a byte length of at least 2. This component cannot complie on this platform."
: : #endif
: : [/code]: :
: :
: : Is this hunky-dori?
[code][/code]
[color=Green]ige> This won't work. "sizoof(...)" operator is evaluated at runtime, and not at compile time. [/color]
[code][/code]
: :
: : thnx
: : {2}rIng
: [code]: [/code]:
: I never thought about that method :-)
: [code]: [/code]:
: I useually do it based off platform.
: [code]: [/code]:
: Or, just use the ANSI [b]stdint.h[/b] :-)
: [code]: [/code]:
: As for undefining a symbol:
: [code]:
: #undef foo
: typedef unsigned int foo;
: [/code]:

Gregry2

: : : [code]: : :
: : : #if sizeof(unsigned short) >= 2
: : : typedef _sint_uint16_ unsigned short
: : :
: : : #elif sizeof(unsigned int) >= 2
: : : typedef _sint_uint16_ unsigned int
: : :
: : : #elif sizeof(unsigned long) >= 2
: : : typedef _sint_uint16_ unsigned long
: : :
: : : #else
: : : #error "No primitive type has a byte length of at least 2. This component cannot complie on this platform."
: : : #endif
: : : [/code]: : :
: : :
: : : Is this hunky-dori?
: [code]: [/code]:
: [color=Green]ige> This won't work. "sizoof(...)" operator is
: evaluated at runtime, and not at compile time. [/color]
: [code]: [/code]:
: : :
: : : thnx
: : : {2}rIng
: : [code]: : [/code]: :
: : I never thought about that method :-)
: : [code]: : [/code]: :
: : I useually do it based off platform.
: : [code]: : [/code]: :
: : Or, just use the ANSI [b]stdint.h[/b] :-)
: : [code]: : [/code]: :
: : As for undefining a symbol:
: : [code]: :
: : #undef foo
: : typedef unsigned int foo;
: : [/code]: :
:
:

[code]
typedef int uint16;
#undef uint16

typedef int uint16;

uint16 a;

int main(){
a=2;

}
[/code]

Then should this compile? I didn't on my computer.
{2}rIng

MT2002

: [code]:
: typedef int uint16;
: #undef uint16
:
: typedef int uint16;
:
: uint16 a;
:
: int main(){
: a=2;
:
: }
: [/code]:
:
: Then should this compile? I didn't on my computer.
: {2}rIng

Interesting... It works fine for me (Visual Studio 2005
and DJGPP GCC).

What compilier do you have?

TurboC doesnt seem to like it, which may emphisize its C++
specific. (Not C)

Gregry2

: : [code]: :
: : typedef int uint16;
: : #undef uint16
: :
: : typedef int uint16;
: :
: : uint16 a;
: :
: : int main(){
: : a=2;
: :
: : }
: : [/code]: :
: :
: : Then should this compile? I didn't on my computer.
: : {2}rIng
:
: Interesting... It works fine for me (Visual Studio 2005
: and DJGPP GCC).
:
: What compilier do you have?
:
: TurboC doesnt seem to like it, which may emphisize its C++
: specific. (Not C)

OH!! THis is a cpp thing...i renamed the file to cpp and it compiled...

I'm doing this component in C...sorry, I probably should of specified that.
{2}rIng

MT2002

: OH!! THis is a cpp thing...i renamed the file to cpp and it
: compiled...
:
: I'm doing this component in C...sorry, I probably should of
: specified that.
: {2}rIng

Ah :-)

If its C, you might need to define the types in the preprocessor:
[code]
#undef int32_t
#undef int64_t

#ifdef _WIN32
# define int32_t int // On Win32, sizeof (int)==32 bits
# define int64_t long int
#elif _WIN64
# define int32_t unsigned short // I think
# define int64_t int
#endif
[/code]
int32_t is guaranteed to be a 32bit on both Win32 and Win64
platforms.

[hr]
[size=1][leftbr].:EvolutionEngine[rightbr][leftbr].:MicroOS Operating System[rightbr][leftbr][link=http://www.mt2002.sitesled.com]Website[/link]
[/size]

ige

: : OH!! THis is a cpp thing...i renamed the file to cpp and it
: : compiled...
: :
: : I'm doing this component in C...sorry, I probably should of
: : specified that.
: : {2}rIng
:
: Ah :-)
:
: If its C, you might need to define the types in the preprocessor:
: [code]:
: #undef int32_t
: #undef int64_t
:
: #ifdef _WIN32
: # define int32_t int // On Win32, sizeof (int)==32 bits
: # define int64_t long int
: #elif _WIN64
: # define int32_t unsigned short // I think
: # define int64_t int
: #endif
: [/code]:
: int32_t is guaranteed to be a 32bit on both Win32 and Win64
: platforms.
:
: [hr]
: [size=1][leftbr].:EvolutionEngine[rightbr][leftbr].:MicroOS
: Operating
: System[rightbr][leftbr][link=http://www.mt2002.sitesled.com]Website[/
: link]
: [/size]

[code][/code]
[color=Green]
ige>
Yes, that (#define) thing is correct.
[code][/code]
But what was this following piece of code supposed to do (this was posted before)?
[code][/code]
undefining a typedef?

[code]: : typedef int uint16;
: : #undef uint16
: :
: : typedef int uint16;
: :
: : uint16 a;
: :
: : int main(){
: : a=2;
: :
: : }[/code][/color]

MT2002

: [code]: [/code]:
: [color=Green]
: ige>
: Yes, that (#define) thing is correct.
: [code]: [/code]:
: But what was this following piece of code supposed to do (this was
: posted before)?
: [code]: [/code]:
: undefining a typedef?
:
: [code]: : : typedef int uint16;
: : : #undef uint16
: : :
: : : typedef int uint16;
: : :
: : : uint16 a;
: : :
: : : int main(){
: : : a=2;
: : :
: : : }[/code]: [/color]
:

It undefines a type, just like its C #define counterpart.
C++ treats typedef as just defined types, hence yes--you can
#undef a typedef. This is C++ only though.

[hr][size=1][leftbr].:EvolutionEngine[rightbr][leftbr].:MicroOS Operating System[rightbr][leftbr][link=http://www.mt2002.sitesled.com]Website[rightbr][/link][/size]