-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Darren Duncan wrote: > The main rationale for the feature I mention is code brevity,
For more context, what language are you writing your code in and how do you get the contents of each column. Using the C api there are 3 ways: 1 - Explicit: "select x,y from foo" (col 0 is x, col 1 is y) 2 - Find name: Use sqlite3_column_name to find "x" and "y" 3 - Brittle: "select * from foo" (breaks on any schema change) > For example, if you have a > 8-field table and you want to return 6 fields, only having to spell out the 2 > you don't want makes for shorter code. But nothing happens if a ninth column is then added. (You'll start getting the column - there won't be errors.) This proposal then becomes more documentation than anything else. A program statically analysing the queries isn't going to be able to infer anything about correctness. To a human it may also look a little scary - using your users/password example, can I tell that your code knows about a recently added 'lmhash' column? [BTW I am a nobody so don't worry too much about this] In summary I understand what you are saying and acknowledge that you feel this would make your code cleaner. However I don't see the benefits being much and would certainly be concerned if something was added to SQLite and then a different syntax became a standard. I would consider myself proved wrong if there are codebases (probably open source) showing that developers have tried to do this sort of thing (eg functions to filter out columns). Roger -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iEYEARECAAYFAkrAYGoACgkQmOOfHg372QSfrQCcDmSWj8c7edneWaGEI1DJwyeD TmgAn08Ygk3phS/XzMdtFZv8TmXx8M7S =WWXW -----END PGP SIGNATURE----- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users