Should I persist this application data?

Question

I have a class in my application that translates path tokens into fully qualified paths. For example: it can take a string like "%MYAPPDATA%" and return C:\Users\user.DOMAIN\AppData\Raoming\MyApp.

Alternatively, the class has an overload to the function that can take an enum instead of a string. For example: it can take the enum AppPaths.MyAppData and return C:\Users\user.DOMAIN\AppData\Raoming\MyApp.

I need to store the "lookup table" somewhere, but I'm not sure what the best method or structure is. Should I use a dataset and write the table to disk? Or just keep in it memory?

A single path value can map to a string and an enum. I suppose I can just keep an array in memory whose index maps to the integer value of the enum and do a search through the array when I'm passed a string.

Thoughts?

Answer 1

Unless it is particularly large, I would just go with the array in memory. For a bit more functionality - and a bit more overhead - you could use a list or a dictionary. For lots of functionality - but lots of overhead - you could use an in-memory dataset.

Again, unless you have more than, say, 250-500 items, it really wouldn't make sense to use disk storage during the operation of your program: the overhead in both latency (retrieval time) and coding is just not worth it.

You may well, of course, keep records on disk as your long-term storage (e.g. to be loaded when the program starts). However, it looks from your problem that you may not even need this if you are pulling the data from the OS.

Answer 2

It all depends on the performance you desire.

You might store the data in database tables (one for the "raw" data, two for the relationship between the keys and the enumeration) and look it up with simple SQL queries. This may not be the fastest way, though, depending on the underlying database and caching mechanism.

You may also use two in-memory maps (dictionaries?) for faster lookup at runtime, intializing them at startup.

Answer 3

I would keep it in memory if it is impossible that it can grow to big - otherwise I'd store it in a database/file for later use.

Of course it depends on your usage. As far as I see you want to store the path to user-specific-data. Could it be that you've got a User-Class where you could store the path of the data?

if(Directory.Exists(User.Current.AppDataPath))

So you could probably get rid of the "lookup table" since every user knows where his AppData is - just a thought...