Should I use bit fields?
There are 3 files:
http://www.mediafire.com/file/16fvmf051l07u6h/bitset.h
http://www.mediafire.com/file/81b547kgb2b7m4y/bitset_detail.h
http://www.mediafire.com/file/7vz6bj02zcn77yb/bitset_test.cpp
I'll post the source here also... please forgive the changes to types/names etc - I was banging my head against a wall trying to see if it was name conflicts - although how it could be when in a unique namespace I don't know!
bitset_detail:
http://www.mediafire.com/file/16fvmf051l07u6h/bitset.h
http://www.mediafire.com/file/81b547kgb2b7m4y/bitset_detail.h
http://www.mediafire.com/file/7vz6bj02zcn77yb/bitset_test.cpp
I'll post the source here also... please forgive the changes to types/names etc - I was banging my head against a wall trying to see if it was name conflicts - although how it could be when in a unique namespace I don't know!
bitset_detail:
|
|
|
|
|
|
EDIT: Actually, I can see the use of set() methods that can set multiple bits at once. For example, if the class' "bits" member was a memory-mapped port, then it would be useful to be able to set or clear or alter the state of multiple bits simultaneously. But
my above lines are wrong... they need disable_if's. Ugh. Here's a corrected example.
First, change the above from functions to enums. Not sure if the compiler will/can elide the function calls, and there is
no need for any of them to be functions anyway.
Next, here is the corrected set() declaration:
my above lines are wrong... they need disable_if's. Ugh. Here's a corrected example.
First, change the above from functions to enums. Not sure if the compiler will/can elide the function calls, and there is
no need for any of them to be functions anyway.
|
|
Next, here is the corrected set() declaration:
|
|
Here is a corrected (and modified) count() function. The old one had a bug.
It shifted 8 bits at a time instead of 4.
It shifted 8 bits at a time instead of 4.
|
|
Can you repost the compile errors with the exact source files you posted?
Unfortunately my boost version predates asio, so I cannot compile your code :(
Unfortunately my boost version predates asio, so I cannot compile your code :(
Using this post: http://www.cplusplus.com/forum/general/27147/page2.html#msg147084
| ~/proj/utils/inc/bitset.h:22: error: expected nested-name-specifier before numeric constant ~/proj/utils/inc/bitset.h:22: error: expected ‘>’ before numeric constant |
|
|
~/proj/utils/inc/bitset.h:35: error: ‘B1’ was not declared in this scope ~/proj/utils/inc/bitset.h:35: error: ‘B2’ was not declared in this scope ... etc etc etc... ~/proj/utils/inc/bitset.h:35: error: ‘B15’ was not declared in this scope ~/proj/utils/inc/bitset.h:35: error: type/value mismatch at argument 1 in template parameter list for ‘template<class T0, class T1, class T2, class T3, class T4, class T5, class T6, class T7, class T8, class T9, class T10, class T11, class T12, class T13, class T14, class T15, class T16, class T17, class T18, class T19> struct boost::mpl::vector’ ~/proj/utils/inc/bitset.h:35: error: expected a type, got ‘0’ ~/proj/utils/inc/bitset.h:35: error: template argument 2 is invalid ~/proj/utils/inc/bitset.h:35: error: template argument 3 is invalid ... etc etc etc... ~/proj/utils/inc/bitset.h:35: error: template argument 16 is invalid |
|
|
Can you try renaming my template parameters from B0 etc to something else like BIT0? Looks kinda like boost.asio might
have already defined the symbol B0, perhaps as a macro?
have already defined the symbol B0, perhaps as a macro?
I extended the mask struct to allow creation of a mask from multiple bit types:
|
|
Do you know of a decent way to convert the types into strings, and then be able to get either a list of strings, or a long concatenated string, of the bits which are set and/or not set?
eg, something like this:
TIA
Steve
eg, something like this:
|
|
TIA
Steve
Last edited on
In namespace detail, add:
In class bitfield, add the public member:
Note: I'm just returning typeid( BitName ).name(), which returns the mangled name. You might consider running
the name through a demangler first. [On gcc, it mangles it by prepending an integer corresponding to the length
of the type, in characters. So for example, the type "bit_one" gets mangled as "7bit_one". I believe there is
a function called __cxa_demangle in cxxabi.h that you might be able to use.]
|
|
In class bitfield, add the public member:
|
|
Note: I'm just returning typeid( BitName ).name(), which returns the mangled name. You might consider running
the name through a demangler first. [On gcc, it mangles it by prepending an integer corresponding to the length
of the type, in characters. So for example, the type "bit_one" gets mangled as "7bit_one". I believe there is
a function called __cxa_demangle in cxxabi.h that you might be able to use.]
jsmith, if I may avail myself upon thee again please!
I'm trying to use your bitset in conjunction with a serialisable message.
Here is some code which I'm trying to use for creating my messages:
Now in my client code, I create a message as such:
And if I now want to use my message, I can do the following:
under the hood, msg_t<bitset_t>::field<int, field0>::operator=(U& d) should set the correct bit in both bitsets mask and changes.
However, even without attempting to instantiate the msg_t template class above (ie: without declaring struct my_msg, compiling fails with the following error:
line 20: error: expected primary-expression before '>' token
Any line where I attempt to use the bitset members fails. Since bitset is a template parameter, I would have thought the compiler would only try to parse it when the template is instantiated.
Is what I'm trying to do possible? Does it make sense?
TIA
Steve
I'm trying to use your bitset in conjunction with a serialisable message.
Here is some code which I'm trying to use for creating my messages:
|
|
Now in my client code, I create a message as such:
|
|
And if I now want to use my message, I can do the following:
|
|
under the hood, msg_t<bitset_t>::field<int, field0>::operator=(U& d) should set the correct bit in both bitsets mask and changes.
However, even without attempting to instantiate the msg_t template class above (ie: without declaring struct my_msg, compiling fails with the following error:
line 20: error: expected primary-expression before '>' token
Any line where I attempt to use the bitset members fails. Since bitset is a template parameter, I would have thought the compiler would only try to parse it when the template is instantiated.
Is what I'm trying to do possible? Does it make sense?
TIA
Steve
Yes, it is possible and it makes sense. And even better, there's an easy answer.
You just have to add the keyword "template" at the right places:
You just have to add the keyword "template" at the right places:
|
|
Last edited on
Wah?!!? what on earth does that mean?! I've never seen syntax like that before?!
Why does the compiler need template there?
Just when I feel I'm getting a grasp on things, the old curve-ball comes along! :)
Thanks once again!
Why does the compiler need template there?
Just when I feel I'm getting a grasp on things, the old curve-ball comes along! :)
Thanks once again!
I'm wondering if Loki's GenScatterHierarchy or something like that would make more sense for what I'm trying to do?
So I looked up in the current C++ standard (n3092). I could quote the text here, but it probably wouldn't help. Anyway,
the section is 14.2 paragraph 4. Basically, the compiler needs it there because the standard says the compiler needs it
there.
the section is 14.2 paragraph 4. Basically, the compiler needs it there because the standard says the compiler needs it
there.
Thanks again - I read up the relevant section - so it's to tell the compiler not to parse the first angle bracket as less-than, but rather as a template argument list.
I have another question relating to this. The below code snippet is slightly different to the previous snippet I posted, but the basic gist is the same.
Given a template class field, which has a member bitset mask (whose type is template param bitset_t), and another template param field_t which specifies the bit position in the bitset, how can I use bitset's local struct bit_pos<field_t>::pos to find the bit position in the bitset?
See code below:
Currently I get an error
line 4: error: expected ‘;’ before ‘::’ token
I have another question relating to this. The below code snippet is slightly different to the previous snippet I posted, but the basic gist is the same.
Given a template class field, which has a member bitset mask (whose type is template param bitset_t), and another template param field_t which specifies the bit position in the bitset, how can I use bitset's local struct bit_pos<field_t>::pos to find the bit position in the bitset?
See code below:
|
|
Currently I get an error
line 4: error: expected ‘;’ before ‘::’ token
Last edited on