Erlang/OTP Forums
Author Message

<  Open Telecom Platform (OTP)  ~  Partial match on a mnesia key column
alamb
Posted: Thu Mar 27, 2008 12:46 pm Reply with quote
User Joined: 26 Aug 2007 Posts: 55 Location: Geneva / Switzerland
Hello,

If I have a mnesia table configured as a set with two columns.

My key column is rather complexe (a tuple with several fields).

If I fetch records using match object (or match key) on a partial match (e.g. {Field1,'_','_',Field4} for example), how is it optimised? Will it use the index which is automatically put on key columns or will it scann the complete table?

Or, should I create an additionnal column with (for example) Field1 duplicated and first fetch on that additional indexed column before filtering the remaining records in memory?

Thanks,

Alex
Back to top View user's profile Send private message
francesco
Posted: Mon Mar 31, 2008 8:31 am Reply with quote
User Joined: 07 Jul 2006 Posts: 249 Location: London
match and match object are optimized on the index fields. If the index fields are not included in the query, all elements are scanned.

Francesco
--
http://www.erlang-consulting.com
Back to top View user's profile Send private message Visit poster's website
alamb
Posted: Mon Mar 31, 2008 10:41 am Reply with quote
User Joined: 26 Aug 2007 Posts: 55 Location: Geneva / Switzerland
But if my table is of type "set", the first column will be indexed automatically?

So there is an index.

If that column is full of tuples such as:

{a,b,c}
{a,x,d}
{b,c,f}
etc...

then I search for {a,b,'–'}

will the index be used. My problem is really: how does mnesia index a column which is made of tuples? Is it actually several indexes? Like first an index on the number of elements of the tuples, then on element one of the tuple, etc... or does it take the whole tuple for example as a string before adding an index?

Thanks,

Alex
Back to top View user's profile Send private message
francesco
Posted: Mon Mar 31, 2008 11:03 am Reply with quote
User Joined: 07 Jul 2006 Posts: 249 Location: London
Quote:
But if my table is of type "set", the first column will be indexed automatically?


The first column is by default the key, used for the hash. For it to be used, all variables have to be set.


Quote:
If that column is full of tuples such as:

{a,b,c}
{a,x,d}
{b,c,f}
etc...

then I search for {a,b,'–'}

will the index be used.


No, it will not, as you are matching on one of the values.

Quote:
My problem is really: how does mnesia index a column which is made of tuples? Is it actually several indexes? Like first an index on the number of elements of the tuples, then on element one of the tuple, etc... or does it take the whole tuple for example as a string before adding an index?


If your element is a tuple, it will not index every element. Setting wild cards in a structure in the index will result every element being compared.

Regards,
Francesco
--
http://www.erlang-consulting.com
Back to top View user's profile Send private message Visit poster's website
alamb
Posted: Mon Mar 31, 2008 11:56 am Reply with quote
User Joined: 26 Aug 2007 Posts: 55 Location: Geneva / Switzerland
Ok, so it means that if my key column is made of tuples such as:

{A,B,C}

and I often will search for {A, something else...}

I ought to add a column to my table containing A (a duplicat of first tuple element) and index on that column.
Back to top View user's profile Send private message

Display posts from previous:  

All times are GMT
View next topic
View previous topic
Page 1 of 1
Erlang Forum - Trap Exit Forum Index  ~  Open Telecom Platform (OTP)
This forum is locked: you cannot post, reply to, or edit topics.

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum
Hosted by Erlang Solutions    |    Powered by phpBB and NoseBleed 1.14