Comments on: Further thoughts on BlitzDB’s Index Handling

By: Toru Maesaka

Toru Maesaka — Sun, 17 Jan 2010 14:36:00 +0000

@Jay Pipes As always, thanks for the awesome advises! This means a lot to me. I was talking to mtaylor last week and it seems MY_COLLATION_HANDLER object inside MY_CHARSET_INFO has lots of goodies inside it. I'll look into these plus your advise and hopefully solve my problem :)

By: Toru Maesaka

Toru Maesaka — Sun, 17 Jan 2010 14:34:06 +0000

@Jobin Augustine Thanks for the encouragement. I think I'll try this approach and see how it goes.

By: Jay Pipes

Jay Pipes — Fri, 15 Jan 2010 17:47:20 +0000

Sorry above should be:

int found= ! my_strcasecmp(…);

the function returns 0 on a match.

=-jay

By: Jay Pipes

Jay Pipes — Fri, 15 Jan 2010 17:03:46 +0000

Hi Toru!

“Perhaps Drizzle/MySQL provides an internal normalization function that respects this. I still need to study this area of the storage subsystem.”

Indeed, there is such a normalization routine :)

my_strcasecmp() is probably what you are looking for. Use it like so:

const char *key= “KEY”;
int found= my_strcasecmp(&my_charset_utf8_general_ci, “key”, key);

if (found)
printf(“found it.”);

You will probably have to also:

#include
#include

Cheers!

-jay

By: Jobin Augustine

Jobin Augustine — Fri, 15 Jan 2010 11:59:24 +0000

I think it is a right approch.
every PK gets a tail :) (index)
otherwise things like range scans will be too expensive.

If you look at other planets (say oracle) every primary key got an index..as they call “unique index”.
if it is named PK, index will have the same name. otherwise system generated name. obviously citizens are not allowed to drop this index.