Right; that's my point -- either 2/3 or 4 is redundant. It's more efficient to do it once, rather than repeat the lower on every file read, which is why I suggested sticking with 4).We currently lowercase in 4 places. I think they are well-defined. 1) The search key(s). This is necessary. 2) On creation of auto database files. This is necessary. 3) On reading a text file (FlatFileStore). This is necessary. 4) On append to the CONFIRM_APPEND file. This is not necessary because of (3). We could remove this, although it's probably called the least frequently of the four.
> But the only way to assume lower() has been done already for the
> auto database cases is to do it ourselves, which you've just suggested
> removing!
All I'm saying is that obviously lines added to a file by TMDA w/ autoappend are going to be lowercased; hence there is no need to re-lowercase when creating an autodb. If the user wants to alter files manually, IMO it's no big deal to require him to conform to that standard. If you're not willing to trust the user that much, we can agree to disagree. But there's nothing invalid with this from a technical standpoint, as opposed to socially.
> The domain portion of an email address is allowed to be in any case
> and the local portion is MTA dependent. The only possible way to
> ensure we match any combination of case is to either lowercase or
> uppercase.
Uh, I wasn't born retarded, and I don't think I suddenly lost 100 IQ points in the last week, either. :/
-Jonathan
_________________________________________________
tmda-workers mailing list ([EMAIL PROTECTED])
http://tmda.net/lists/listinfo/tmda-workers
