[Exolent[jNr]]

Originally Posted by Lulu the hero What you are arguing about are CPU cycles.

I know. That was the purpose of me stating to use &31 in the first place.

[Bugsy]

IMO, an include file for this type of thing is overkill. A thread containing the few macros (which already exists) would suffice.

#define player_to_flag(%1)		(1<<(%1-1))

#define player_to_flag(%1)		(1<<(%1 & 31))

Using ( id & 31 ) instead of ( id - 1 ) is more efficient and will simply change when a ( 1 << 0 ) shift is used.

id - 1:

PHP Code:

1 << 0 //player_to_flag(1)
1 << 1 //player_to_flag(2)
...
1 << 30 //player_to_flag(31)
1 << 31 //player_to_flag(32) 


id & 31:

PHP Code:

1 << 1 //player_to_flag(1)
1 << 2 //player_to_flag(2)
...
1 << 31 //player_to_flag(31)
1 << 0 //player_to_flag(32) 


[Exolent[jNr]]

Originally Posted by Bugsy Using ( id & 31 ) instead of ( id - 1 ) is more efficient and will simply change when a ( 1 << 0 ) shift is used.

I had already pointed that out but he seemed stubborn to change.

[Bugsy]

Originally Posted by Exolent[jNr] I had already pointed that out but he seemed stubborn to change.

Yea I know, I just wanted to show him the difference between the two.

[Lulu the hero]

Done, thanks for the tips, guys. I hope it will be usable. Please if you can, check it on a 64 bit server.

[Exolent[jNr]]

Originally Posted by Lulu the hero Done, thanks for the tips, guys. I hope it will be usable. Please if you can, check it on a 64 bit server.

These macros are very simple and don't need to be tested.

[Bugsy]

#if cellbits==32
	#define cellbytes	4
	#define flagmask	31
#else
	#define cellbytes	8
	#define flagmask	63
#endif

// Convert's id to a flag
#define player_to_flag(%1)		(1<<(%1&flagmask))

The number of players supported by HL1 will always be 32, regardless if the server is running on a 32 or 64 bit system. There is no need to change the mask for 32 & 64 bit systems if only 32 bits are going to be used in both cases. It would be best to make a separate macro using ( 1 << %1 ) for 64-bit systems and continue to use ( 1 << ( %1 & 31 ) ) for 32-bit.

#if cellbits==32
	#define cellbytes	4
	#define player_to_flag(%1)	(1<<(%1&31))
#else
	#define cellbytes	8
	#define player_to_flag(%1)	(1<<%1)
#endi

[Lulu the hero]

Done. What other functions should be inserted to the include file, that you think is related to the ones that are already in? I have a few ideas:
- convert a number to an any given dimension string( we currently have num_to_binstr and num_to_hexstr, but how about a something like: num_to_custom( num, dest[], len, dimension )? )
- get/set/clear/toggle a SPECIFIED bit in an integer, not only ID( would almost be the same )
- ROL, ROR and alike functions( ROL= ROtate Left and ROL = ROtate Right )
- Some kind of comparison for two booleans


Edit:
Added ROL and ROR stocks...

[Lulu the hero]

Added toggle_flag / clear_flag / set_flag / is_flag_set macros to make the include file usable for not just only players.

Warning! Got some bugs in num_to_bin/hexstr. I'm sorting it out now, gonna upload working version ASAP.

Edit : debugged it, works fine.

[Lulu the hero]

Originally Posted by fysiks I don't see how your include has anything to do with booleans. ...

True. Renamed the include file to simply: bit

Corrected the num_to_hexstr ( I've forgot, that 0xF is 15, which is 0b1111, which is only 4 bits long, so it is not a byte, but a nibble ).

Added the zero_fill parameter to num_to_bin/hexstr functions( while removed the length parameter, do we really need that? ).

Added stocks:
- get_highest(integer, count_type) // count_type = COUNT_BIT(1) | COUNT_NIBBLE(4) - gives back the place of the most significant bit/nibble
- count_bits(integer) // counts the 1 bits in an integer