ROM : Running the server : RT ASCII stuff...

Gammon Software Solutions forum

See www.mushclient.com/spam for dealing with forum spam. Please read the MUSHclient FAQ!

Entire forum
ROM
Running the server
RT ASCII stuff...

Home \| Users \| Search \| FAQ
Username:	Register forum user name
Password:	Forgotten password?

RT ASCII stuff...

Reply to this subject Start a new subject Refresh page

Pages: 1 2

Posted by

Nash USA (20 posts)

Date

Wed 26 May 2004 02:14 AM (UTC) [ quote ]

Message

ROM2.4b6, and I'd assume most of its siblings, uses 32 bit ascii rt blah something to assign bits to various things, like act_ and aff_ etc. The problem is you only can have 32 different things like this for each, if I wanted to expand that system to like 64 or something, where exactly would I go... or rather is there anything I should be concerned about in taking this approach so that I can add more act and aff types to a mud?

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #1 on Wed 26 May 2004 02:30 AM (UTC) [ quote ]

Message

32 bit ascii rt huh??

I think what you mean is that it uses a 32-bit integer with bit masks... and sine the integer is 32 bits in size you only have 32 bits to toggle on and off.

The easy way to expand this is to use long longs (__int64 on Windows, long long on Unix), which will give you 64... but, you'll have to make an awful lot of changes. Notably, any code that accesses these will have to be aware of the new 64-bit version.

Optionally you could look at the later versions of SMAUG, which have something called EXT_BV I believe which gets around this problem.

Of course, the right thing to do would be to make your own data type, but that's a considerably harder task. :)

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #2 on Thu 27 May 2004 06:01 AM (UTC) [ quote ]

Amended on Thu 27 May 2004 06:05 AM (UTC) by Nash

Message

I probably changed the ints to long the bad way, because I get a large large error message and it all starts with:

merc.h:47: conflicting types for `system'
/usr/include/stdlib.h:694: previous declaration of `system'

and by ascii rt I meant..

/* RT ASCII conversions -- used so we can have letters in this file */

#define A 1
#define B 2
etc..

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #3 on Thu 27 May 2004 04:10 PM (UTC) [ quote ]

Message

Quote:
I probably changed the ints to long the bad way, because I get a large large error message and it all starts with:

You probably changed way more ints to long than you were supposed to.

Quote:
and by ascii rt I meant..

/* RT ASCII conversions -- used so we can have letters in this file */

#define A 1
#define B 2
etc..

...and these have what to do with the bitvectors? Those are just letters.

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #4 on Fri 28 May 2004 04:22 AM (UTC) [ quote ]

Message

What specific integers do I need to change into longs for this to work then, or... am I going about it wrong..?

Those are used to recognize the bits from the save files, those are assigned to act_ flags or aff_flags or any of the non-number based flaggies.. likkkkeeeeeeeee

#define ACT_MAGE (R)
#define ACT_THIEF (S)
#define ACT_FIGHTER (T)
#define ACT_NOALIGN (U)
#define ACT_NOPURGE (V)
#define ACT_OUTDOORS (W)
#define ACT_INDOORS (Y)
#define ACT_IS_HEALER (aa)
#define ACT_GAIN (bb)
#define ACT_UPDATE_ALWAYS (cc)

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #5 on Fri 28 May 2004 04:29 AM (UTC) [ quote ]

Message

Then there must be some other conversion going on, because to make it work as flags - if indeed they are using the bitvector scheme and there is no reason they wouldn't be. There must be a bitshift somewhere where those values are set. Can you give an example of where those flags are actually set?

The flags actually *are* number-based, by the way... that's what the defines are all about.

You have to change the variables which store these flags and the routines that read from them, but frankly it's a rather complicated coding task.

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #6 on Fri 28 May 2004 04:58 AM (UTC) [ quote ]

Message

these are like flags... like, they just exist, and your mob or players can have them, like a thief would be flagged with 'act_thief' and it can just run through and check if the character has the act flag 'act_thief' I believe that code in my last post was it being defined, because... like (ee) is like "#define ee 1073741824"

so yes, I know they're number based..
#define ACT_THIEF (S)
#define S 262144
so act_thief = 262144, which is part of some 32 bit saving system ROM2.4b6 uses..
'ee' is the largest it can go to, any attempt to expand it crashes with the error that I've exceeded the limit for an integer, which is true! Since integers are tiny..

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #7 on Fri 28 May 2004 05:35 AM (UTC) [ quote ]

Message

*ahem* I think I know how a bit-based flag system works. :) I'm not sure why they found it necessary to give the bitmasks letters... but, whatever.

Ints aren't as small as you think. I think you're not quite sure how these flags work. They are bitmasks. The number you gave just happens to be, in binary: 1000000000000000000. And if you look at R, lo and behold, it will be: 100000000000000000. Every time your letters move down (silly notation, BVxx or BMxx is much better) you will lose one zero in binary.

When you combine flags, for instance: 1 (1) and 2 (10), you get in binary 11.
So, if you were to combine from A to S, you get in binary: 1111111111111111111, which is 524,287 in decimal.

The maximum value for integers is 4294967296, assuming unsigned integers. You're probably using signed integers, in which case you "only" have 2,147,483,648. If 2 billion is "tiny" for you, I wonder what big is! That being said, the numeric decimal size of the integer is basically worthless; here we are only interested in the bits.

In any case you have two options. Either you stick with a bitmask system and go from 32-bit integers to 64-bit integers (long long on Unix, __int64 on Windows)... or you get rid of the bitmask system, and use something else (an array of bools, for example, or even smarter, a custom data structure that uses an array of integer bitmasks.) In the bitmask system, you are *always* limited to the number of bits you have available. 32-bit integers will provide no more than 32 flags. That's just the way it is. And if you use 64-bit integers, you will have no more than 64.

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #8 on Fri 28 May 2004 02:35 PM (UTC) [ quote ]

Amended on Fri 28 May 2004 02:38 PM (UTC) by Nash

Message

I know how small integers are, just... they're used as 32 ones and zeroes to flag stuff in this case, so no matter what I'm limited to 32, but yes, it'd be nice to make a datastructure, but I really haven't the slightest clue how, and I probably won't be expanding it so much that I need more than 64 bits, but I apparently don't know which integers I wanna convert to long, and if it's just as easy as changing the variable from 'int whatever;' to 'long whatever;', but if ya have any explanation of applying a data structure to the various fields where it currently uses the bit-ness, I'd like to know..

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #9 on Fri 28 May 2004 07:46 PM (UTC) [ quote ]

Message

I said 'long long', not 'long'. If you change to long you likely won't be changing anything since on many systems ints and longs are both of 32 bits.

You need to change:
- the variables that store the flags
- the macroes that read/write the flags
- the functions that save/load the flags to file

Proceed at your own risk, however :)

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #10 on Sat 29 May 2004 06:13 AM (UTC) [ quote ]

Message

Am I doing something wrong to define a longlong? It's my understanding you do simply 'long long <variable>;', it works except I get the following errornicities...

db.c:2068: warning: decimal constant is so large that it is unsigned
->if (IS_SET(mob->act,ACT_SHAMAN))<-

handler.c:2681: warning: decimal constant is so large that it is unsigned
->if (act_flags & ACT_SHAMAN ) strcat(buf, " shaman");<-

tables.c:106: warning: decimal constant is so large that it is unsigned
->{ "shaman", ff, TRUE },<-

and ACT_SHAMAN is set as 'ff' which is set as '2147483648'

I set the bit variable in act_flags to a long long, but it apparently don't like it... and... just wondering if I define longlongs wrong

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #11 on Sat 29 May 2004 06:58 AM (UTC) [ quote ]

Message

Well, for one, you need to change all functions that access the new format.

For two, typically you don't define these constants the way your MUD is... you define them as bitshifts. e.g. 1 << 32

That should remove the warnings, but I suggest you look at code that actually uses that technique e.g. SMAUG 1.4.

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #12 on Sun 30 May 2004 12:50 AM (UTC) [ quote ]

Amended on Sun 30 May 2004 12:55 AM (UTC) by Nash

Message

hm... lost me on the bitshift :D

the closest thing found there was "#define BV31 (1 << 31)" which isn't used in the flags, apparently they use a table for the various flags and such... but I'm still learning to code-alize, and that's way way beyond me..

Posted by

David Haley USA (3,881 posts)

bio Moderator

Date

Reply #13 on Sun 30 May 2004 01:02 AM (UTC) [ quote ]

Message

"Code-alize"? :P

The BVxx is the way you're "supposed" to do the flags, i.e. proper bitshifts and not decimal constants.

Do you understand binary and how bitmasks work?

If I may say so, no offense but you're trying something rather difficult but don't seem to know much about coding. More likely than not you'll just get frustrated here. Have you looked into the extended bitvector system that SMAUG has like I suggested?

David Haley aka Ksilyan
Head Programmer,
Legends of the Darkstone

http://david.the-haleys.org

Posted by

Nash USA (20 posts)

Date

Reply #14 on Sun 30 May 2004 03:32 AM (UTC) [ quote ]

Amended on Sun 30 May 2004 03:37 AM (UTC) by Nash

Message

I do understand bitmasks and binary, and I have coding knowledge, but I've been more of a logic programmer, and I'm porting my knowledge of other languages over to C, which is a transistion that's not being very easy for me to make.. I learn best by asking a lot of questions and this is just something I don't understand too well, sorry for bugging you for so much clarification. I don't think I totally understand it, but it did work, and it's whittled the errors down to when you save, it seems to print weird stuff

->Afg† 0 0 0<-
(and the weird holy cross looks like an a with a circle ontop of it in shell..)

It seems to save new mobs funny... that mob had the flags 'npc' and 'shaman', and printed the above... the code for saving is in olc_save.c and I think the following function...

char *fwrite_flag( long long flags, char buf[] )
{
char offset;
char *cp;

buf[0] = '\0';

if ( flags == 0 )
{
strcpy( buf, "0" );
return buf;
}

/* 64 -- number of bits in a long long */

for ( offset = 0, cp = buf; offset < 64; offset++ )
if ( flags & ( (long)1 << offset ) )
{
if ( offset <= 'Z' - 'A' )
*(cp++) = 'A' + offset;
else
*(cp++) = 'a' + offset - ( 'Z' - 'A' + 1 );
}

*cp = '\0';

return buf;
}

and it's read in by this..

long long fread_flag( FILE *fp)
{
int number;
char c;
bool negative = FALSE;

do
{
c = getc(fp);
}
while ( isspace(c));

if (c == '-')
{
negative = TRUE;
c = getc(fp);
}

number = 0;

if (!isdigit(c))
{
while (('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z'))
{
number += flag_convert(c);
c = getc(fp);
}
}

while (isdigit(c))
{
number = number * 10 + c - '0';
c = getc(fp);
}

if (c == '|')
number += fread_flag(fp);

else if ( c != ' ')
ungetc(c,fp);

if (negative)
return -1 * number;

return number;
}

this function also kinda stuck out to me...

long long fread_flag( FILE *fp)
{
int number;
char c;
bool negative = FALSE;

do
{
c = getc(fp);
}
while ( isspace(c));

if (c == '-')
{
negative = TRUE;
c = getc(fp);
}

number = 0;

if (!isdigit(c))
{
while (('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z'))
{
number += flag_convert(c);
c = getc(fp);
}
}

while (isdigit(c))
{
number = number * 10 + c - '0';
c = getc(fp);
}

if (c == '|')
number += fread_flag(fp);

else if ( c != ' ')
ungetc(c,fp);

if (negative)
return -1 * number;

return number;
}

The dates and times for posts above are shown in Universal Co-ordinated Time (UTC).

To show them in your local time you can join the forum, and then set the 'time correction' field in your profile to the number of hours difference between your location and UTC time.

5,136 views.

This is page 1, subject is 2 pages long: 1 2

Reply to this subject Start a new subject Refresh page

Go to topic: Search the forum

Comments to: Gammon Software support
Forum RSS feed ( http://www.gammon.com.au/rss/forum.xml )