FishEye and Crucible

In LoadAuctionImage, if Auc-ScanData cannot be loaded, it creates a new image table in (global) AucScanData, and saves a copy in (global save) AucAdvancedData.
Due to a typo this is not reloaded in the next session, and stays stuck in the file indefinitely. (AucAdvancedData.Scandata/AucAdvancedData.ScanData)

We should not be trying to store an entire (uncompressed) image table in Auc-Advanced save file - this is what Auc-ScanData is designed to do.

Possible options if Auc-ScanData cannot be loaded:

A) Create an empty image each session; don't save it between sessions. Essentially like 'deleting' the scandata every session.

B) Restrict to storing a single Auctionhouse; warn the user before overwriting if they scan a different Auctionhouse than the one currently saved.

C) As (A), but also prevent any scanning to avoid Stats being affected. Otherwise if you scan, reload & scan again, each auction on the Auctionhouse will be 'seen' twice by Stat-Simple, Stat-Histo, Stat-StdDev, etc.

D) Refinement of (C): Allow the scan, but block ScanProcessor messages. This allows the user to get a scan image for competition, but doesn't affect Stats.

For "of the" items XXX of the bear, XX of illusion etc. If they have a negative suffix (the number before uniqueID) then the unique ID has the scaling factor built in to it.
So XX of the bear with 855555 unique ID may be +64 intel +50 sta but unique ID 77777 would be +50 intel +90 sta (All values made up)

See Scaled Random Suffixes part of the link you provided.

Basically we could extract just the suffix value for storage but the space saved is too trivial for the snatch list. Only when you are storing 1000's of itemstrings ala Beancounter do the savings add up enough.