Igor Tandetnik wrote: > Martin Engelschalk > <engelsch...@codeswift.com> wrote: > >> The problem appeared when my users eliminated leading blanks from the >> data. >> > > I'm not sure I understand. What does this have to do with collating > numbers? Are you saying you were storing numbers as strings, with > leading blanks, and had a collation function to actually compare them as > numbers? Why? > My collation does not compare the data as numbers but interpreted them in a different way. Comes from a bad data design :-) > >> Is it a good idea that the actual type of the data influences whether >> it is compared with the collaction function or not? >> > > Well, collation functions as they exist today take char* or wchar_t* as > a parameter, and no type information. How would you pass a number to it? > Maybe it would have been wise to design them to accept sqlite3_value* > pointers as parameters, the way custom functions do. But that wasn't > done, so it is what it is. > The way the data is passed to the function is not a problem for me, but the fact that the function is not called if all the data is numerical. > If for whatever reason you do want to compare numbers as strings, you > can write > > order by Cast(NumberColumn as string) collate MyCollation > This is a very good idea! It might just give me back my weekend! > Igor Tandetnik > > > > _______________________________________________ > sqlite-users mailing list > sqlite-users@sqlite.org > http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users > >
_______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
