Ensure that Item.folder works in all cases.

[hoffie]

The internal attribute Item._folder should always be initialized,
at least with None.
If possible, _folder should be pre-populated with parent data from
the outside as this should be better performing than doing a repeated lookup.

Fixes #31.

[jelly]

I agree that it is a bug, but I think we should fix it either in the init() or add a check in the Item.folder() code to check if the store exists. So that Item creation with a mapiobj is a bit less buggy.

The last solution also reveals the following bug

AttributeError: 'Item' object has no attribute 'store'

So I guess we then need to set the store explicitly in the init().

[hoffie]

Thanks for the fast reply. I am a bit unsure whether you are actually suggesting the same what I proposed or something different. :)

To clarify, I would suggest to initialize _folder in __init__() (or as part of the class definition) as Item.folder() is just one possible place for errors due to this behaviour. Future code might introduce similar bugs.

In addition, I only propose to pre-initialize new Item objects with a parent if possible as this should help with performance.

[srepmub]

'self' is a Store, so not really the direct parent of the item.. but note that 'Item.folder' looks up the correct parent folder using PR_PARENT_ENTRYID.

[srepmub]

alright, jelle, so I guess we want to merge the Item(parent=self) changes when Item() is called from a Folder method. and probably it should be Item(folder=self) by now.

[hoffie]

@srepmub, yes, one Item(folder=self) instance seems to be wrong as you noted.
From your comments I assume that this repository is out of date compared to ZCP 7.2? So your primary repository is located somewhere internally? My assumption was that the GitHub version was newer and 7.2 just contained a stable snapshot it of.

Is there anything else I should do with the pull request?

[jelly]

Looks good, we will merge this when the repo is up to date