Re: [Monetdb-developers] about MAL optimize

11 Aug 2010


      what I think is right? the ideas and intension of MAL optimizers contain
three parts: 1. programming language/compiler optimization 2. databose query
optimization 3. some optimization only for column-oriented database but not
for row-oriented database,such as mitosis,mergtable,dataflow etc.as a whole,
all MAL optimizers is unque for MonetDB. Do I understand right?

在 2010年8月11日 下午5:21，kun ren 写道：
...
Thanks very much, you means the ideas and intension of MAL optimizers
contain three parts: 1. programming language/compiler optimization 2.
databose query optimization 3. some optimization only for column-oriented
database but not for row-oriented database,such as
mitosis,mergtable,dataflow etc
  Best regards!
2010/8/11 Stefan Manegold 
...
The ideas / intension of some MAL optimizers, such as alias removal, dead
code elimination, (to some extend) join ordering, etc.  are generic and
traditional, both in database query optimization and progarmming language
/
compiler optimization.  The genral optimizer design and architecture
combining (to some extend) both database query optimization and
progarmming
language / compiler optimization, as well as some of the optimizers (e.g.,
mitosis, mergtable, dataflow, etc.), and obviously the very implementation
of all MAL optimizers, is very unique to MonetDB5/SQL and MAL.
Stefan
...
Thanks，professor Stefan Manegold,when I use explain,I get the optimizer
MAL
plan.
 I have another question, so we know that many classic optimizer for
row-oriented database，and can I say the MAL optimizer is unique for
column-oriented database such as MonetDB? or some of MAL optimizer is
unique
for column-oriented?
Best regards!
2010/8/11 Stefan Manegold 
...
On Wed, Aug 11, 2010 at 04:23:42PM +0800, kun ren wrote:
...
I read sql_optimizer setting in monetdb5.conf ,and default setting
is:
sql_optimizer=default_pipe
default_pipe=inline,remap,evaluate,costModel,coercions,emptySet,aliases,mitosis,mergetable,deadcode,commonTerms,joinPath,reorder,deadcode,reduce,garbageCollector,dataflow,history,multiplex
...
...
so there are many optimizer,but why when i debug MonetDB,and dump
...
...
...
my  sql statement,but i can't find the optimizer mal ,and  don't
enter
of
the
...
optimizer? like there are no optimizer,
if you prefix your SQL query with EXPLAIN to see the MAL plan, you get
...
...
optimizer MAL plan, i.e., all optimizers have already been called and
done
their work.
Stefan
...
2010/8/11 Stefan Manegold 
...
Yes, if you write MAL "by hand" (or have some program create it)
...
...
does
...
...
explicitly call a MAL optimizer, that MAL optimizer is indeed used
to
optimize the MAL code according to its very purpose.
Stefan
On Wed, Aug 11, 2010 at 03:49:47PM +0800, kun ren wrote:
> It is only used by SQL? and how to designation the optimiser？but
when
I
> write MAL,for example,
> i:=0;
> b:= "done";
> barrier go:= true;
> c:=i+1;
> d:="step";
> v:=d;
> io.print(v);
> i:=c;
> redo go:= i<2;
> exit go;
> io.print(b);
> optimizer.aliasRemoval();
>
> the last optimizer.aliasRemoval() is used for optimizing?
>
> 2010/8/11 Fabian Groffen 
>
> >  On 11-08-2010 15:25:15 +0800, kun ren wrote:
> > >    Recently,I am intersting in MAL optimize，and have read
...
...
manual
...
...
of
> > the
> > > MAL optimize, but it is so long, Possible if we want to use
mal
optimize
> > ,we
> > > must  degignation the optimize  consideration,  such as
alias
removal,
> > > constant expression,join path optimizer;And then I dedug the
MonetDB,but
> > > default it don't use any optimize.so what do you think about
On Wed, Aug 11, 2010 at 04:38:01PM +0800, kun ren wrote:
the mal
the
that
the
the
...
...
mal
...
...
> > > optimize?
> >
> > It is only used when you use SQL.  The sql_optimizer setting
in
> > monetdb5.conf controls which optimisers are run, which by
default
are
> > quite a few.
> >
> >
> >

...
...
...
...
> > This SF.net email is sponsored by
> >
> > Make an app they can't live without
> > Enter the BlackBerry Developer Challenge
> > http://p.sf.net/sfu/RIM-dev2dev
> > _______________________________________________
> > Monetdb-developers mailing list
> > Monetdb-developers@lists.sourceforge.net
> >
https://lists.sourceforge.net/lists/listinfo/monetdb-developers
> >
>

...
...
...
...
> This SF.net email is sponsored by
>
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev
> _______________________________________________
> Monetdb-developers mailing list
> Monetdb-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/monetdb-developers
--
| Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl |
| CWI,  P.O.Box 94079 | http://www.cwi.nl/~manegold/  |
| 1090 GB Amsterdam   | Tel.: +31 (20) 592-4212       |
| The Netherlands     | Fax : +31 (20) 592-4199       |
--
 | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl |
| CWI,  P.O.Box 94079 | http://www.cwi.nl/~manegold/  |
| 1090 GB Amsterdam   | Tel.: +31 (20) 592-4212       |
| The Netherlands     | Fax : +31 (20) 592-4199       |
--
 | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl |
| CWI,  P.O.Box 94079 | http://www.cwi.nl/~manegold/  |
| 1090 GB Amsterdam   | Tel.: +31 (20) 592-4212       |
| The Netherlands     | Fax : +31 (20) 592-4199       |