Re: flags in a bitmap

On 2025-07-04 16:57, Ruud H.G. van Tol wrote:
>
> On 2025-07-04 16:16, Ruud H.G. van Tol wrote:
>> [switched from user to dev]
>>
>> On 2025-07-04 15:51, Bill Allombert wrote:
>>> On Fri, Jul 04, 2025 at 12:38:59PM +0200, Ruud H.G. van Tol wrote:
>>>>> ? install(bitset,vWL)
>>>>> ? x=2^(2^20);
>>>>> ? forprime(p=2,2^20,bitset(~x,p))
>>>>>     ***   last result computed in 8 ms.
>>>>> ? my(S=0);for(i=0,2^20-1,S+=bittest(x,i));S
>>>>> %16 = 82025
>>>>>     ***   last result computed in 159 ms.
>>>> That indeed looks good, and I can test it.
>>> Great, I have made a GIT branch bill-bitset
>>>
>>> commit 8632eaf7528554234aaa90d9a503476452cbd736 (HEAD -> 
>>> bill-bitset, origin/bill-bitset, master)
>>> Author: Bill Allombert <[email protected]>
>>> Date:   Thu Jul 3 23:57:00 2025 +0200
>>>
>>>      New GP functions bitset, bitclear
>>
>> Thanks. First (minor) remark:
>>
>>
>> -Help: bitclear(~n,i): clear bit i of n (in place)
>> +Help: bitclear(~x,n): clear bit n of x (in place)
>>
>> -Help: bitset(~n,i): set bit i of n in place
>> +Help: bitset(~x,n): set bit n of x in place
>>
>>
>> to align it with bittest, and with the code.
>
> And of course another:
>
> diff --git a/src/headers/paridecl.h b/src/headers/paridecl.h
> +void    bitclear(GEN x, long i);
> +void    bitset(GEN x, long i);
>  long    bittest(GEN x, long n);
>
> Those long i's might also want to be n's.
>

These 3 are less equally worded than I would expect:

-Help: bitclear(~x,n): clear bit n of x (in place)
=Help: bitflip(~x,n): flip bit n of x in place, assuming x >= 2^n.
-Help: bitset(~x,n): set bit n of x in place, assuming x>=2^n.

+Help: bitclear(~x,n): clear bit n of x in place, assuming x >= 2^n.
=Help: bitflip(~x,n): flip bit n of x in place, assuming x >= 2^n.
+Help: bitset(~x,n): set bit n of x in place, assuming x >= 2^n.


This looks special to me:

? my(x=2^4); print(x); bitclear(x,4); print(x);
16
Vecsmall([])

? my(x=2^4); print(x); bitflip(x,4); print(x);
16
Vecsmall([])

as I expected 0.

-- Ruud