Date: Thu, 12 Mar 2020 12:52:57 -0700
Ramy -
You're saying that using strings to initialize bitsets is infeasible
because of their size. To me, this sounds like you're concerned about, in
this case, 80 bytes of RAM?
You go on to suggest that the bitset string constructor should accept
another argument, indicating the base of the string, and that the to_stirng
member function should accept a base.
But this also infeasible, albeit for a different reason: you're now
suggesting that the standard library should be able to perform an
arbitrary-precision conversion of a string from one base to another.
That would cost a lot more than 80 bytes of code... and std::bitset is an
odd place to put that conversion operation.
I believe that, for your needs, it should be easy to write a pair of
functions:
template<size_t N>
std::bitset<N> bitset_from_hex(std::string_view);
template<size_t N>
std::string hex_from_bitset(const std::bitset<N>&);
Does that work for you?
-- Jorg
On Thu, Mar 12, 2020 at 12:25 PM ramy koudsi via Std-Proposals <
std-proposals_at_[hidden]> wrote:
> Apologies, I must have been referring to an old spec by accident.
> C++11 and newer does indeed accept a string in one of the constructors.
>
> However, the accepted string must be zeroes and ones. This doesn't seem
> very feasible, especially as we get larger bitsets.
> Using current C++11 or newer, to initialize a bitset with a value of 80
> bits, we would have to pass an 80 character string.
>
> For this reason I see a need for the bitset string constructor to accept
> another argument, indicating the base of the string.
> This way, initializing a value can be done with a much smaller string.
>
> A base can also be passed to the to_string member function, telling it to
> output a string in that base, instead of just zeroes and ones.
>
> Thank you,
>
> Ramy
>
>
> On March 12, 2020 at 6:07 AM, Magnus Fromreide <magfr_at_[hidden]>
> wrote:
>
> On Thu, Mar 12, 2020 at 01:49:42AM -0400, ramy koudsi via Std-Proposals
> wrote:
>
> Hi,
>
>
> std::bitset does not have a size limit, however it has an initialization
> size limit. Bitsets cannot currently be initialized with anything bigger
> than a long long int.
>
> To remove this limitation I propose adding another constructor that
> accepts a std::string object.
>
>
> std::bitset already (since c++11) have a constructor accepting a
> std::basic_string.
>
> /MF
>
> I see its use like:
>
> std::bitset<80> bigbs(“FFFF95D3FFFFAC27FFFF”);
>
>
> The example above uses hex strings, but can easily be changed to accept
> strings of any base by adding another constructor accepting a string and an
> argument showing the base.
>
> std::bitset<12> littlebs(“AFD”, 16);
>
>
> In addition, a member function to_string(int base) would be nice.
>
> std::string littlestr = littlebs.to_string(16);
>
> // littlestr is “AFD”
>
>
> I wanted to hear your thoughts about this. Has anyone thought of this and
> come up with a better solution?
>
>
> Thank you
>
>
> Ramy
>
> --
>
> Std-Proposals mailing list
>
> Std-Proposals_at_[hidden]
>
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>
> --
> Std-Proposals mailing list
> Std-Proposals_at_[hidden]
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>
You're saying that using strings to initialize bitsets is infeasible
because of their size. To me, this sounds like you're concerned about, in
this case, 80 bytes of RAM?
You go on to suggest that the bitset string constructor should accept
another argument, indicating the base of the string, and that the to_stirng
member function should accept a base.
But this also infeasible, albeit for a different reason: you're now
suggesting that the standard library should be able to perform an
arbitrary-precision conversion of a string from one base to another.
That would cost a lot more than 80 bytes of code... and std::bitset is an
odd place to put that conversion operation.
I believe that, for your needs, it should be easy to write a pair of
functions:
template<size_t N>
std::bitset<N> bitset_from_hex(std::string_view);
template<size_t N>
std::string hex_from_bitset(const std::bitset<N>&);
Does that work for you?
-- Jorg
On Thu, Mar 12, 2020 at 12:25 PM ramy koudsi via Std-Proposals <
std-proposals_at_[hidden]> wrote:
> Apologies, I must have been referring to an old spec by accident.
> C++11 and newer does indeed accept a string in one of the constructors.
>
> However, the accepted string must be zeroes and ones. This doesn't seem
> very feasible, especially as we get larger bitsets.
> Using current C++11 or newer, to initialize a bitset with a value of 80
> bits, we would have to pass an 80 character string.
>
> For this reason I see a need for the bitset string constructor to accept
> another argument, indicating the base of the string.
> This way, initializing a value can be done with a much smaller string.
>
> A base can also be passed to the to_string member function, telling it to
> output a string in that base, instead of just zeroes and ones.
>
> Thank you,
>
> Ramy
>
>
> On March 12, 2020 at 6:07 AM, Magnus Fromreide <magfr_at_[hidden]>
> wrote:
>
> On Thu, Mar 12, 2020 at 01:49:42AM -0400, ramy koudsi via Std-Proposals
> wrote:
>
> Hi,
>
>
> std::bitset does not have a size limit, however it has an initialization
> size limit. Bitsets cannot currently be initialized with anything bigger
> than a long long int.
>
> To remove this limitation I propose adding another constructor that
> accepts a std::string object.
>
>
> std::bitset already (since c++11) have a constructor accepting a
> std::basic_string.
>
> /MF
>
> I see its use like:
>
> std::bitset<80> bigbs(“FFFF95D3FFFFAC27FFFF”);
>
>
> The example above uses hex strings, but can easily be changed to accept
> strings of any base by adding another constructor accepting a string and an
> argument showing the base.
>
> std::bitset<12> littlebs(“AFD”, 16);
>
>
> In addition, a member function to_string(int base) would be nice.
>
> std::string littlestr = littlebs.to_string(16);
>
> // littlestr is “AFD”
>
>
> I wanted to hear your thoughts about this. Has anyone thought of this and
> come up with a better solution?
>
>
> Thank you
>
>
> Ramy
>
> --
>
> Std-Proposals mailing list
>
> Std-Proposals_at_[hidden]
>
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>
> --
> Std-Proposals mailing list
> Std-Proposals_at_[hidden]
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals
>
Received on 2020-03-12 14:55:56