What would be suitable SL container(s) for frequent search, sort etc but infrequent insertions after it has once been initialized? The container(s) would hold unique, custom objects and I'd like to understand the principles behind any recommendation as well. Thanks
ps: google not very helpful: Google
There's not a lot of choice.
A sorted vector (or array) that you binary search, giving logarithmic time for access.
A set (or map), which are implemented as red-black trees (or something similar), giving logarithmic time for access.
If you didn't need sorting, then an unordered_set (or unordered_map), which are implemented as hash tables, giving constant time for access.
If you give the details of your object and how often you insert after initialization then someone might be able to suggest a more specialized data structure.
[QUOTE]ps: google not very helpful: Google
Then perhaps you need to change your search parameters? Remember that if one "search" doesn't produce the desired results you need to refine your search parameters. Perhaps something like C++ selecting proper container might produce better results.
Jim
algorism: thanks for your reply
the objects will be stocks (perhaps, later, other financial securities) where each stock object has a unique securities number. there is no requirement for sorting the original container but we'll be running screens on this container, selecting stocks on the basis of various criteria and the selected stocks (much smaller in number than the original
container size) will need sorting within their own containers but that's another issue.
Insertions (and deletions) would normally happen quarterly when equity indices usually rebalance unless there's interim corporate events like an IPO (an insertion), takeover/merger (a deletion) etc.
Another thing I should mention is that sizeof(object) will probably be quite large as it'll have a several data members representing underlying company fundamentals, behavior of its stock price, daily trading volumes, etc but this should not have any effect upon the choice of the container per se? From your list I think the unordered containers might be more up my street
jim: i posted one of my searches' link but admittedly you defined the search terms better. further down in your link there's a C++11 version of the container cheat-sheet that also seems to point out the unordered containers for this exercise. Thanks
If by "quarterly" you mean the equivalent of "every designated three months in a year", then it sounds like you're looking for a persistent database rather than an in-memory container, with in-memory containers used only after accessing the database to retrieve the relevant records (hence you will not have a huge number of large objects in-memory all at once).Originally Posted by sean_cantab
Look up a C++ Reference and learn How To Ask Questions The Smart Way
hi laserlight: the stock selection screens can run directly upon the memory container but for deleting objects can we do it directly from the memory container (here I'm thinking primarily .txt files) or do we need to read entire memory container to an in-memory container first? for insertions, yes, they can be written directly to the memory container. thanks
