MESSAGE
| DATE | 2011-05-19 |
| FROM | Ruben Safir
|
| SUBJECT | Re: [NYLXS - HANGOUT] C++ Workshop - creating a distribution list
|
The key function in creating a distribution list in out project is the
function that exists in the stats namespace, but which is not a member
of the Distribution class:
template
void mount_individual_data_point(chainlist::List * tabulate,
chainlist::List > * table){
T val;
stats::Distribution * j;
val = *(tabulate->cursor()->value()); //get a value
table->cursor()= table->front(); //check to see if the
//distribution list exists
if(!table->cursor()){ // if not add a distribution table to the
//List of distributions
j = new stats::Distribution (val);
table->insert(*j ); //now we have at least one
delete j;
j=table->cursor()->value();//and increased its population
j->increase_occ();
}else{
//otherwise search for a distribution node described as
//value
table->find_value(val);
if( table->cursor() ){
j=table->cursor()->value();//and increase its population
j->increase_occ();
}else{//otherwise add a new node
j = new stats::Distribution (val);
table->insert( *j ); //now we have one for that value
delete j;
j=table->cursor()->value();//and increased its population
j->increase_occ();
}
}
}
A detailed look at this templated function is valuable. The purpose of
stats:: mount_individual_data_point(chainlist::List *, chainlist::List *) is
to take two lists, one of type T, which in our working example
translates to integers, and to walk through it creating or growing a
second List of type stats::Distribution. As the function walks
through the first list, it hunts for unique new samples, and tallys
their population in the list in the List::Distribution. Buy sending
both lists by pointer, we avoid uneeded copy overhead.
it starts be declaring in int val, which is just for later readability
and convience, which holds the current value that we are examining.
values in our List class are pointers, so we copy by value the
deferenced value into val.
we then create a pointer to a stats::Distribution object which we use to
construct new Distribution objects and destroy them as needed,
preventing memory leaks.
stats::Distribution * j;
This just then uses the List's cursor to find where in the list we
want to analyze. This is set by the calling function. as
mount_individual_data_point only works on a single member of the
population at a time.
val = *(tabulate->cursor()->value()); //get a value
we then check to see if our Distribution list is populated. It is up to
the user of this function to assure that the Distribution List is
actually allocated and not just a pointer, otherwise the infix operation
will fail:
table->cursor()= table->front();
If cursor is equal to front, then we haven't done anything to the list
yet, it is an empty shell
if(!table->cursor()){ // if not add a distribution table to the List of distributions
j = new stats::Distribution (val);
table->insert(*j ); //now we have at least one
delete j;
j=table->cursor()->value();//and increased its population
j->increase_occ();
Keep in mind that cursor will equal zero if pointed at front, if no
members have yet been added, and so we start by creating a new
Distrubution object pointed to (j) with value val, insert it in our
List > * table, passing j by value, and then clean up by
deleting j.
Finally we pass the address of the new internal Distribution Node within
table back to (j), which is a pointer, and call the Distribution member
increase_occ() which incrememnt the population tally represented in the
private data as int occurances of the Distribution class.
Now if it is not a fresh List< Distribution >, then we need to
search to see if out Distribution node already exists. We do this by
leveraging our chainlist::List member find_value() which sets the cursor
of the list to the first occurance of a value. Since the List<
Distribution > has all unique members, we only need to find the first
occurance. If the value doesn't exist in the List, then the function
set the cursor to NULL
}else{
//otherwise search for a distribution node described as
//value
table->find_value(val);
if( table->cursor() ){
j=table->cursor()->value();//and increase its population
j->increase_occ();
}else{//otherwise add a new node
j = new stats::Distribution (val);
table->insert( *j ); //now we have one for that value
delete j;
j=table->cursor()->value();//and increased its population
j->increase_occ();
}
}
}
Any questions?
Ruben
|
|
