How does my set class have duplicate values?

Question

I have a Set which appears to have duplicate values. Here is the pseudocode:

list<SOBject__c> list1 = [select id from SObject__c where blah blah blah];
list<SObject__c> list2 = [select id from SObject__c where blah blah blah];
list<SObject__c> list3 = [select id from SObject__c where blah blah blah];

set<SObject__c> combinedSet = new set<SObject__c>();
combinedSet.addall(list1);
combinedSet.addall(list2);
combinedSet.addall(list3);

A few of the items in list1 and list3 are identical. My expectation is that when I add all three lists to the set, that addall() will ignore any element that's already present in the set. However, when I iterate through combinedSet and look at SObject__c.Name, the items that overlap in list1 and list3 are in the set twice. (I do not have records with the same SObject__c.Name.)

How is this possible?

EDIT: Since there have been questions, I decided to be a bit more verbose and share most of the actual code. I'm going to share the part of the code that I believe is relevant. I have an object called Show__c. I'm querying against Show__c three times with three different conditions. That gives me three lists. But there are some records in Show__c for which two of the conditions are true, so two of the three lists overlap. That's why I use a set to remove the duplicates.

list<Show__c> list1 = [select id from Show__c where condition A];
list<Show__c> list2 = [select id from Show__c where condition B];
list<Show__c> list3 = [select id from Show__c where condition C];
set<Show__c> allShows = new set<Show__c>();
allShows.addall(list1);
allShows.addall(list2);
allShows.addall(list3);

Next I have another object called Hop__c which has two fields, a Master-Detail with Show__c and a Date. I insert the Show__c records from the set into Hop__c.

list<Hop__c> toInsert = new list<Hop__c>();

for (Show__c s : allShows) {
    Hop__c b = new Hop__c();
    b.Show__c = s.id;
    b.Week_Starting__c = date.today().adddays(1); 
    toInsert.add(b);
}

if (!toInsert.isempty()) insert toInsert;

After this runs (this is a Scheduled Apex), I look in Hop__c (which was previously empty) by generating a "Shows with Hops" report from within the SFDC UI. In the report I see the several instances of the same Show__c, which should not happen. When I run each individual query that generates the lists, I see some of the same records in list1 and list3, and those same records show up as duplicates in Hop__c. It's as if the set is storing duplicates.

Answer 1

Unless you've selected every possible field in the <sObject> with each query, I think you need to do something like limit the returned fields to only the same 2 fields or simply put your results into sets of Ids as I'm showing in the code below. Once you do, I believe you'll get different results than what you got initially:

map<Id, sObject> mapSO1 = new map<Id,sObject>([select Id,Name Where condition A]);
map<Id, sObject> mapSO2 = new map<Id,sObject>([select Id,Name Where condition B]); 
map<Id, sObject> mapSO3 = new map<Id,sObject>([select Id,Name Where condition C]);

set<Id> combinedSet = new set<Id>();

combinedSet.addall(mapSo1.keyset());
combinedSet.addall(mapSo2.keyset());
combinedSet.addall(mapSo3.keyset());

Now, test the above records for duplicates using the method of your choice in a for loop.

Without knowing the conditions and the fields you've returned, it's difficult to say why you're getting the results you are since you're currently putting all the records into sets as sObjects.

I don't believe an instance of a list of sObject records containing records with say 2 fields in it is the same as an instance of the same list where each record contains 5 fields that have been returned from the same query. If evaluated, I believe they wouldn't show as being equal or equivalent to one another.

However, the lists would both contain the same record Ids and the resulting sets would only contain one record Id. To me, the latter would be the "litmus test"; thus the reason for changing the structure of your test.

Using sObjects, it would seem the sets could contain two records with the same Id since instances of the two sObject records would not be equivalent because of the difference in the number of fields that each contained. Merging instance records with the same Id would be the only way to make them equivalent to where they contained all the same data and thus were "equal".