[spambayes-dev] Re: Pickle vs DB inconsistencies

[Meyer, Tony]

> >    <script> -d spambayes.db    # load a DBM store
> >    <script> -D spambayes.pkl   # load a pickle store
> 
> No, that is definitely the *worse* option, because it has no 
> mnemonic value.  (I'm ignoring historical precedent, user 
> expectations, and all that.)  That's why I like -d to load 
> a DB[M] file and -p to open a pickle.

Fair enough.  Hopefully most users are using the config file method to
set this information anyway, and so wouldn't even be effected by a
change.

Since everyone else has been relatively quiet about this, I suppose you
could just go ahead and check in the changes when you want.

[Dropping pickles]
> My recent experiences with bugs in the DB[M] storage -- still 
> the ostensible subject of this thread! -- are a good argument 
> against this. Tim P's words were along the lines of, "The 
> pickle implementation is so simple that it's obviously 
> correct", and it's good to have a gold standard, even if it 
> is a memory hog and slow to startup.

Very true.  I believe his (TimS) reasons are to do with the reasonably
complex message information database / statistics database / master
database business (half implemented in message.py).  In any case, I'll
let him push his own agenda :)

> Hey, I'm suggesting a whole alternative naming scheme: 
> sb-<verb>, where <verb> is "train", "score", "filter", 
> "export", "import", etc.

Are you suggesting that the source files are renamed, or that the setup
renames them?  Does cvs allow for renaming without losing the history?

=Tony Meyer