[APBeta] DSS Image library

Paul Rodman paul at ilanga.com
Mon Mar 16 17:46:35 PDT 2009


Folks,

Currently, in V1 and V2, the DSS/user image library consists of a  
whole mess of image files, other files containing info pertinent to  
those images, and an image index, which is really just a database of  
file paths.

This way of doing things has/had the advantage of being flexible,  
friendly to incremental backups, etc. However, it's also slow, and has  
a lot of spaghetti code surrounding it which is buggy and difficult to  
maintain/extend. I'm getting a lot of bugs reported related to this  
feature.

What I propose is to create a single-file database of the images and  
attendant information. This would have the advantages of much faster  
access, less chance of corruption, smaller and more maintainable code,  
and easier extension to support new sources, etc.

The disadvantages would be be (a) those of any large single file, such  
as backing up, etc., a single corruption damaging the entire  
database*, etc., and (b) non-backwards-compatibility with V1.6.

I don't see (b) as a problem going forward.

The size of the file would be approximately the size of the DSS Image  
Cache folder now, so you wouldn't be saving/losing disk space.

I would probably read the current cache to create the initial  
database, so you wouldn't have to download all those pesky images again

What do you people think about this? Please air your views here or  
privately with me.

Paul R.

* I would be using the SQLite database format, which is very robust  
and efficient - in both speed and use of disk space.



More information about the APBeta mailing list