From the NannyMUD documentation
LAST CHANGE
2000-12-23
TOPIC
NAME
class1 - A log from TMI, part 1.
DESCRIPTION
This is part 1 of some logs from TMI, The MUD Institute, a MUD
dedicated to teaching LPC. TMI has gone down since, and been replaced
by TMI-II.
Profezzorn says: ok, we go to mappings
Profezzorn says: theory first, or examples?
Tlofgren says: theory first, then examples. :)
Profezzorn says: ok, mappings are ascosiative arrays
Profezzorn says: if that is how it is spelled
Tlofgren says: probably not, but I'm with ya
Bwilliams nods solemnly.
Profezzorn says: the idea is to make an array that can be indexed on
anything
Profezzorn says: in lpc the type is called mapping, and it is declared
like:
Profezzorn says: mapping map;
Profezzorn says: it is indexed just as an array:
Profezzorn says: map[index]
Tlofgren nods solemnly.
Profezzorn says: where the index can be of any type
Profezzorn says: when assigning to an index in a mapping
Profezzorn says: map[index]=data
Profezzorn says: the index will be added to the mapping if it doesn't
exist already
Profezzorn says: so a mappign can grow when more data is assigned to it
Bwilliams says: there is no allocate required for mappings?
Profezzorn says: well, you have to init the array, but the size isn't
important
Profezzorn says: the mapping I mean
Profezzorn says: you init the mapping to an empty mapping like this:
Profezzorn says: map=([]);
Bwilliams nods solemnly.
Profezzorn says: that is also how you write constant mappings
Khan says: the question i have: how slow an memory expensiv is an
mapping against an usual lets say mixed * ????
Profezzorn says: mapings are slower and use more memory, but the exact
values depends on the implementation
Khan says: hmm
Profezzorn says: amylaar mudos and 3.1.2 all have different
implementations
Bwilliams nods solemnly.
Profezzorn says: the way to write constant mappings is this:
Profezzorn says: ([index1:data1,index2:data2,index3:data3 .... ])
Profezzorn says: so if if I wrote
map=([index1:data1,index2:data2,index3:data3]);
Profezzorn says: and then map[index1] I would get data1
Bwilliams says: is undefinedp(map["index"]) the standard on all
drivers?
Profezzorn says: no
Profezzorn says: that is mudos specific I think
Bwilliams nods solemnly.
Bwilliams says: how do you check for existance of an index?
Profezzorn says: when an index isn't present in a mapping, it normally
returns 0
Khan nods solemnly.
Profezzorn says: in mudos you can examine that 0 closer with
unefinedp() too see if it was a zero from a data in
the mapping, or a zero because the index wasn't there
Profezzorn says: any questions so far?
Khan says: no ..
Tlofgren shakes and quivers like a bowlful of jelly.
Bwilliams says: or use member_array(keys(map),"index") == -1
Profezzorn says: keys is also mudos-specific
Bwilliams sighs deeply.
Bwilliams nods solemnly.
Tlofgren says: what's w/ the different drivers?
Tlofgren says: how different are they?
Profezzorn says: in some areas they are very different
Bwilliams says: that is a real topic!
Profezzorn nods solemnly.
Profezzorn says: however, the more 'standardized' way to get an array
of all indexes in an array is m_indices(map)
Tlofgren says: ok, carry on then
Khan says: hmm on 3.2.1... u could use m_indices(map) ??!
Khan grins evilly.
Profezzorn says: and m_values(map) to get an array of the values
Bwilliams says: and on amylaar?
Profezzorn says: amyaar use m_indices / m_values too
Bwilliams says: where mudos uses keys() and values()
Khan says: we have amylaar 03.02.1@22.284 ..
Khan nods solemnly.
Profezzorn says: mudos has m_indices and m_values too, as aliases for
keys and values
Profezzorn says: it is vital to know that the arrays returned by
m_indices / m_values is indexed equally
Bwilliams says: but somewhat randomly in MudOS
Profezzorn says: that m_values(map)[e]=map[m_indices(map)[e]] for all
e)
Profezzorn says: it is vital to remember that m_delete isn't
destructive in all drivers
Profezzorn says: on 3.1.2 it just returns a mapping with the index
deleted
Profezzorn says: so you have to write map=m_delete(map,index);
Profezzorn says: any questions?
Bwilliams shakes and quivers like a bowlful of jelly.
Profezzorn says: ok, let's move on to mkmapping()
Tlofgren nods solemnly.
Jngan nods solemnly.
Profezzorn says: mkmaping does the opposite of m_indices and m_values
Profezzorn says: ie. map is always equal to
mkmapping(m_indices(map),m_values(map))
Bwilliams nods solemnly.
Profezzorn says: ie. it creates a mapping from an array of index and
an array of values
Tlofgren nods solemnly.
Profezzorn says: then there are only 3 standard efuns connected to
mappings left
Profezzorn says: mappingp(X) which returns 1 if X is a mapping
Profezzorn says: and filter_mapping / map_mapping
Profezzorn says: filter_mapping(map,function_name,object to call
function in,optional extra argument)
Profezzorn says: that's the approximate syntad for filter_mapping
Profezzorn says: if I write it like:
Bwilliams says: like filter_array()
Profezzorn says: filter_mapping(map,fun,ob,extra)
Profezzorn says: it will be easier to describe
Profezzorn says: yes, it works just like filter_array
Bwilliams nods solemnly.
Profezzorn says: it calls ob->fun(x,extra) for every value in the
mapping with x set to that value
Khan says: sorry .. got to go
Khan waves good-bye.
Bwilliams waves good-bye.
Khan says: bye
Profezzorn waves good-bye.
Khan just left the school.
Tlofgren says: can you give an example of when to use filter_mapping?
Tlofgren says: I don't really see the use at this time.
Profezzorn says: hmm
Profezzorn says: the use isn't very broad
Tlofgren nods solemnly.
Profezzorn says: but, let's say that you have a mapping which contains:
Profezzorn says: ([ player_name : level ])
Bwilliams says: oh, checking exits[] for existance of loaded rooms
Profezzorn says: ie. it is indexed on the name, and contains the level
Profezzorn says: and you want to extract all wizards
Tlofgren nods solemnly.
Profezzorn says: then you use filter mapping
Tlofgren says: ah, ok
Profezzorn says: filter_mapping and map_mapping has one big
disadvantage though
Profezzorn says: they don't send the index (in this case the name) to
the function
Profezzorn says: as you might have guessed, map_mapping works exactly
like filter_mapping, but it returns a mapping where
the values are those returned from the function
Profezzorn says: hope you don't have problems with linebreak...
Profezzorn smiles happily.
Bwilliams says: I'm ffine
Tlofgren says: I don't.
Profezzorn says: any questions?
Bwilliams says: what do they pass to the function if not the index?
Profezzorn says: the value
Bwilliams says: so you test the value for inclusion in the set,
returning 1/0?
Profezzorn says: yes
Profezzorn says: in filter_array that is
Profezzorn says: or do you mean the set of indexes
Bwilliams says: the filter function returns true/false if the value is
acceptable for the filter, right?
Profezzorn says: yes
Bwilliams says: what does map_mapping return?
Profezzorn says: map_mapping(map,fun,ob) does the same as
mkmapping(m_indices(map),map_array(m_values(map),fun,ob))
Bwilliams says: /map_array/filter_mapping/ ?
Ktaylor waves good-bye.
Profezzorn says: are you asking if they are exchengable?
Ktaylor just left the school.
Bwilliams nods solemnly.
Bwilliams sighs deeply.
Profezzorn says: no, not really
Profezzorn says: map_mapping(map,fun,ob) does the same as
mkmapping(m_indices(map),map_array(m_values(map),fun,ob)),
but faster
Bwilliams nods solemnly.
Profezzorn says: ok, as an example, I'll show a filter_mapping I just
coded in lpc
Profezzorn says: varargs mapping filter_array(mapping m,
string fun,
object o,
mixed extra)
Profezzorn says: {
Profezzorn says: mapping map;
Profezzorn says: int e;
Profezzorn says: mixed *a,*b;
Profezzorn says:
Profezzorn says: map=([]);
Profezzorn says: a=m_values(m);
Profezzorn says: b=m_values(m);
Profezzorn says: for(e=0;efun must do a lookup of the
index rather than operating on the value
Bwilliams says: or a and b are the same array!
Profezzorn says: I see that, a should be m_indices
Profezzorn says: not m_values
Bwilliams grins evilly.
Tlofgren nods solemnly.
Profezzorn says: here's another example
Bwilliams says: and map[b[e]]=a[e] is reversed
Profezzorn says: you're right :)
Profezzorn says: I should write this in advance, but it's hard when
you don't know who's coming...
Profezzorn says: :)
Jngan smiles happily.
Tlofgren nods solemnly.
Profezzorn says: /* consider a shop that may sell 10 items / reset */
Profezzorn says: mappign map;
Profezzorn says: #define MAX 10
Profezzorn says: void reset(int arg)
Profezzorn says: {
Profezzorn says: if(!arg) return;
Profezzorn says: map=([]);
Profezzorn says: }
Profezzorn says: int sell(string what)
Profezzorn says: {
Profezzorn says: string name;
Profezzorn says: /* some code to figure out things... */
Profezzorn says: name=(string)this_player()->query_real_name();
Profezzorn says: if(map[name]>MAX)
Profezzorn says: {
Profezzorn says: write("Come back later.\n");
Profezzorn says: return 1;
Profezzorn says: }
Profezzorn says: map[name]++
Profezzorn says: /* rest of sell code */
Profezzorn says: }
Tlofgren nods solemnly.
Jngan nods solemnly.
Profezzorn unloads the projector.
Profezzorn puts something on the projector.
Profezzorn says: another missing ; :)
Profezzorn says: on line 20
Tlofgren nods solemnly.
Bwilliams says: 1s/mappign/mapping/ :)
Profezzorn nods solemnly.
Profezzorn says: any questions on this example?
Tlofgren says: nope. :)
Bwilliams says: that map[name]++ is a cute trick
Tlofgren says: I agree.... useful
Jngan nods solemnly.
Profezzorn says: on some muds map[name]++ bugs a little, so writing
map[name]=map[name]+1 might be wiser
Profezzorn says: but that's just old muds
Profezzorn says: the bug is that it doesn't become 1 if the index
doesn't exist, it becomes 2
Profezzorn says: anyway, I would like to say something about mappings
and +
Profezzorn says: mappings can be added with +
Profezzorn says: which is very useful from time to time BUT
Profezzorn says: indexes in mappings doesn't have to be unique
Profezzorn says: when several data have the same index there is no way
of knowing which of those data you get by writing
map[index]
Bwilliams says: and you get only one?
Profezzorn says: yes
Profezzorn says: yes
Profezzorn says: oops
Bwilliams smiles happily.
Tlofgren says: I think it's odd that they don't have to be unique...
can't see the use for having similar indexes...
Profezzorn says: map[index]=data always keeps the indexes unique, as it
assings over old data
Tlofgren nods solemnly.
Profezzorn says: but mkmapping() and + doesn't check that the indexes
are unique, they just do what they are told
Tlofgren nods solemnly.
Jngan nods solemnly.
Bwilliams says: but map1+map2 can create duplicates
Profezzorn says: yes
Tlofgren says: yeah, true.
Profezzorn says: if the same index is present in both map1 and map2,
the result will have two such index
Profezzorn says: this differs a bit between the drivers though
Profezzorn says: I think dgd checks and always make the indexes unique
Jngan says: got an example?
Bwilliams says: I know MudOS doesn't
Profezzorn says: I know 3.1.2 doesn't
Profezzorn says: anyway, I think we have covered how mappings
_should_ work now
Bwilliams nods solemnly.
Profezzorn says: in the advanced course I will cover how 3.1.2 mappings
can be used if you know what you're doing :)
Profezzorn says: ie. implementation-specific stuff
Profezzorn says: one last word though
Profezzorn says: anothe prime use of mappings is to do binsort-like
things
Profezzorn says: like if you want all objects with the same short()
Tlofgren nods solemnly.
Bwilliams snaps his fingers.
Profezzorn says: then you loop over the array of objects, call short,
and put them into a mapping
Profezzorn says: as we are out of time, how about doing that for
homework?
Profezzorn grins evilly.
Bwilliams grins evilly.
Tlofgren says: heh
Jngan giggles inanely.
Tlofgren says: yeah, see you in a month. :P
Profezzorn says: any homework submitted will get comments on efficiency
and coding style
Profezzorn smiles happily.
Bwilliams says: how do we submit homework?
Tlofgren says: ok, thanks for the class. =)
Profezzorn says: dunno yet :)
Profezzorn says: we'll see about that next week
Tlofgren waves good-bye.
Profezzorn says: I'll try to have some examples on regexp() ready then
Bwilliams nods solemnly.
Jngan nods solemnly.
Tlofgren says: gotta go to sleep. c ya
Bwilliams says: I'll try to be here then....thanks for the info
Jngan says: is there a place we can ftp for the log or something like
that?
Profezzorn says: not yet