Re: Maximum JOINED tables

2002-06-07 Thread Roger Baklund

* Benjamin Pflugmann
 On Thu 2002-06-06 at 19:17:10 +0200, [EMAIL PROTECTED] wrote:
 [...]

  But... is this the only feature of mysql which is dependant on the
  processor architecture?

 See it the other way: You are not garantueed more than 31 tables in a
 join. If you happen to have a 64Bit architecture, you can use more.

Yes... a positive attitude makes life easier. :)

However, I think my question is valid... I don't think it is a good idea to
limit the number of joins based on the processor architecture, and this
makes me wonder what other 'strange' limitations I might find in mysql...
some questions to Victoria/MySQL AB:

mysql is considered to be platform independant, and I suppose MySQL AB is
trying to keep it that way...?

What (other) features of mysql are dependant on the processor architecture?

Will the max number of joined tables (on 32-bit platform) stay at 31 in
version 4.x, or is it planned to increase this limit?

Is this limitation by design, and if so, why? (I suspect it is by
accident, and that's ok. That would mean it will probably be fixed in the
future.)

  I'm kind of nervous about this, because I'm currently working on a 'very
  normalized' database schema, and it is possible I will need more than 31
  tables in a single select... including multiple joins to the
  same tables, of
  course. It would be very bad if my design turned out to be 64-bit
  dependant...

 AFAIK, if you need so many tables, you get a major problem: MySQL
 figures out the best join order by trying all combinations. This
 becomes increasingly slow as you are going to join more tables. In
 other words: Trying to use 30 tables in one select probably will be
 too slow as that it would matter whether so many tables are allowed in
 a join or not.

hm...

The join order optimization is done before the query is executed, so I guess
you are not talking about the _query_ beeing to slow, but the _preparation_
of the query... right?

How slow are we talking here...? Normally the join order is decided within
0.001 sec... (I'm only guessing) When you say slow, do you mean 0.01, or 0.1
or 1.0 sec or what?

In our current implementation of this system, we have about 20 tables in the
main select statement, and we have never seen this problem. Will it be a
problem with 24 tables? 28 tables? You probably don't know the answer to
these questions, but maybe you can tell me more about why you suspect 30
tables would be too slow...?

 Of course, you could work around it by using STRAIGHT_JOIN, but I
 assume it is to cumbersome to figure out the optimal join order for
 all SELECTs.

I was kind of hoping mysql would do that for me... but I guess I could use a
combination, using STRAIGHT_JOIN for some of the joins.

I can split my query in two (or more), but I hope I don't have to, because I
suspect this will slow down the performance, and I need to always have a
result within 0.5 sec...

--
Roger
sql


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Re: Maximum JOINED tables

2002-06-07 Thread Victoria Reznichenko

Roger,
Friday, June 07, 2002, 12:43:38 PM, you wrote:

RB * Benjamin Pflugmann
 On Thu 2002-06-06 at 19:17:10 +0200, [EMAIL PROTECTED] wrote:
 [...]

  But... is this the only feature of mysql which is dependant on the
  processor architecture?

 See it the other way: You are not garantueed more than 31 tables in a
 join. If you happen to have a 64Bit architecture, you can use more.

RB Yes... a positive attitude makes life easier. :)

RB However, I think my question is valid... I don't think it is a good idea to
RB limit the number of joins based on the processor architecture, and this
RB makes me wonder what other 'strange' limitations I might find in mysql...
RB some questions to Victoria/MySQL AB:

RB mysql is considered to be platform independant, and I suppose MySQL AB is
RB trying to keep it that way...?

Roger, MySQL is platform independent. It works on any platform with
the same data, but if the architecture is 64 bit, you can use it for more
power than on 32 bit.

RB Will the max number of joined tables (on 32-bit platform) stay at 31 in
RB version 4.x, or is it planned to increase this limit?

I don't know.

RB Is this limitation by design, and if so, why? (I suspect it is by
RB accident, and that's ok. That would mean it will probably be fixed in the
RB future.)

You can change it (set 63) in 32 bit CPU by hack in compilation, but
making join map 64 bit on 32 bit box makes the work of MySQL slower.




-- 
For technical support contracts, goto https://order.mysql.com/?ref=ensita
This email is sponsored by Ensita.net http://www.ensita.net/
   __  ___ ___   __
  /  |/  /_ __/ __/ __ \/ /Victoria Reznichenko
 / /|_/ / // /\ \/ /_/ / /__   [EMAIL PROTECTED]
/_/  /_/\_, /___/\___\_\___/   MySQL AB / Ensita.net
   ___/   www.mysql.com




-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-07 Thread Roger Baklund

* Victoria Reznichenko
 RB mysql is considered to be platform independant, and I suppose
 RB MySQL AB is trying to keep it that way...?

 Roger, MySQL is platform independent.

I did not say it was not... :)

 It works on any platform with
 the same data, but if the architecture is 64 bit, you can use it for more
 power than on 32 bit.

If the feature set is different on different platforms, the product is not
truly platform independant. The number of joins in a query is a part of the
feature set, IMHO.

It is fully possible that right now someone is developing an application on
a 64bit platform, not knowing that the system will not work on 32bit
platforms simply because they have too many joins.

It is not the 'lack of power' on 32bit platforms that will prevent it from
working, it's the _platform dependant_ limitation in the number of joins.

Ok, _now_ I am saying that mysql is not platform independant, but of course,
this 'incompatibility' is not something most users will run into, so I guess
we could say it is 99% true that mysql is platform independant, or that
mysql is 99% platform independant... :)

(There are also issues with letter casing in table names and different
implementations of the floating point methods.)

 You can change it (set 63) in 32 bit CPU by hack in compilation, but
 making join map 64 bit on 32 bit box makes the work of MySQL slower.

And in this case 'slower' means...? (If we are talking about milliseconds
per query, I don't see the problem. If we are talking about milliseconds per
traversed row, there would be a problem.)

Has anyone ever done this hack, and if so, was the result usefull?

Thank you for the info, Victoria! :)

...and sorry to those who got fooled by me pointing at crashme telling them
the limit was 64 in 3.23.x...

--
Roger
sql


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-07 Thread Benjamin Pflugmann

Hi.

On Fri 2002-06-07 at 11:43:38 +0200, [EMAIL PROTECTED] wrote:
[...]
 However, I think my question is valid... I don't think it is a good idea to
 limit the number of joins based on the processor architecture, and this
 makes me wonder what other 'strange' limitations I might find in mysql...
 some questions to Victoria/MySQL AB:

Well, of course, they could limit the join to 31 tables on all
platforms.

But then people would start to ask, why this limit is artificially low
on 64 bit platforms.

 What (other) features of mysql are dependant on the processor
 architecture?

file size (indirectly... it is dependend on the OS, but the OS is
often dependend on the architecture).

I can not think of any other, currently.

[...]
 Is this limitation by design, and if so, why? (I suspect it is by
 accident, and that's ok. That would mean it will probably be fixed in the
 future.)

I can only guess, but I think it is still there, because increasing it
would be fruitless as long as finding the optimal join is in O(2^N).

Additionally, people seem to seldom need so many tables in a single
join, and that's probably, too, why the join order optimizer is still
the way it is.

 The join order optimization is done before the query is executed, so
 I guess you are not talking about the _query_ beeing to slow, but
 the _preparation_ of the query... right?

I am not sure what your point is.

The client will have to wait too long, no matter where the time is spent.

 In our current implementation of this system, we have about 20 tables in the
 main select statement, and we have never seen this problem. Will it be a
 problem with 24 tables? 28 tables? You probably don't know the answer to
 these questions, but maybe you can tell me more about why you suspect 30
 tables would be too slow...?

Because people had problems with that in the past and asked on the
list for the reason and the answer (by Monty, I think) was that the
join order optimization takes the additional time (in comparison to a
STRAIGHT_JOIN). The delay was in the order of seconds, IIRC.

You'd have to try out. Since the time is doubling with each table,
let's say 31 tables would need 100 secs, with 20 tables you would
hardly notice it, as it would be only 1/2048 of that, i.e. less than
0.05 secs.

You see why allowing 64 tables futile? Presuming processing one order
would be as simply as one plain instruction, the join order optimizer
would need on a Pentium III 500 (225.4 MIPS):

#tables
 10  0.045 secs
 20  0.0047 secs
 30  4.76 secs
 40  1.35 hours
 50 57.81 days
 60162.20 years

Of course, real scaling will differ as pure MIPS don't show real
speed. But you get the idea.

For your purpose, you have to try out, if speed decrease with 30
tables matters to you. Please post any result you find.

Bye,

Benjamin.

-- 
[EMAIL PROTECTED]

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-07 Thread Roger Baklund

* Benjamin Pflugmann
 On Fri 2002-06-07 at 11:43:38 +0200, [EMAIL PROTECTED] wrote:
 [...]
  However, I think my question is valid... I don't think it is a
 good idea to
  limit the number of joins based on the processor architecture, and this
  makes me wonder what other 'strange' limitations I might find
 in mysql...
  some questions to Victoria/MySQL AB:

 Well, of course, they could limit the join to 31 tables on all
 platforms.

 But then people would start to ask, why this limit is artificially low
 on 64 bit platforms.

Well... :)

I would agree to this if any 64 bit based functionality was limited to 64
bit platforms, but that is not the case. For instance, we can use BIGINT on
all platforms. I don't think 'people' would think there is a connection
between the processor word width and the number of joins in a single
select... but of course, I could be wrong.

I do not suggest that the limit should be lowered to 31 on all platforms,
but I find it hard to believe that it is impossible to implement joins in a
different way, so that the max number of joins would be the same on all
platforms. Monty  company have of course worked hard to prevent platform
dependency, and it seems as this is the only part of mysql directly
dependant on the processor architecture. I find it strange, I don't say it
is not true.

  What (other) features of mysql are dependant on the processor
  architecture?

 file size (indirectly... it is dependend on the OS, but the OS is
 often dependend on the architecture).

 I can not think of any other, currently.

If file size qualifies, then the table name letter casing issue and the
differences in the floating point functions also qualifies. But all of these
are OS dependant, not directly processor architecture dependant. And I think
the number of joins in a query is a _feature_ on a different level.

 [...]
  Is this limitation by design, and if so, why? (I suspect it is by
  accident, and that's ok. That would mean it will probably be
 fixed in the
  future.)

 I can only guess, but I think it is still there, because increasing it
 would be fruitless as long as finding the optimal join is in O(2^N).

You are probably right.

 Additionally, people seem to seldom need so many tables in a single
 join, and that's probably, too, why the join order optimizer is still
 the way it is.

Yes... and seldom = sometimes... :)

I don't know the limits for any of the other rdbms, but crashme reports +64
for IBM, Microsoft, Oracle and PostgreSQL.

  The join order optimization is done before the query is executed, so
  I guess you are not talking about the _query_ beeing to slow, but
  the _preparation_ of the query... right?

 I am not sure what your point is.

Just making sure we are talking about the same thing. :)

 The client will have to wait too long, no matter where the time is spent.

The nature of my application allows for a relatively slow query preparation,
because I have few simultaneous queries. I guess up to 0.2 sec could have
been acceptable. (I want the answer within 0.5 sec.)

Query execution time will always increase for each join you add to a select,
but that was of course _not_ what you was talking about. I should have known
better. :)

  In our current implementation of this system, we have about 20
 tables in the
  main select statement, and we have never seen this problem. Will it be a
  problem with 24 tables? 28 tables? You probably don't know the answer to
  these questions, but maybe you can tell me more about why you suspect 30
  tables would be too slow...?

 Because people had problems with that in the past and asked on the
 list for the reason and the answer (by Monty, I think) was that the
 join order optimization takes the additional time (in comparison to a
 STRAIGHT_JOIN). The delay was in the order of seconds, IIRC.

Ah. Thanks.

 You'd have to try out. Since the time is doubling with each table,
 let's say 31 tables would need 100 secs, with 20 tables you would
 hardly notice it, as it would be only 1/2048 of that, i.e. less than
 0.05 secs.

Yes, I see. (31-20=11, 2048 = 11 bits)

 You see why allowing 64 tables futile?

Well... apparently it _is_ allowed on 64 bit platforms... :)

But I understand why implementing it for 32 bit platform is not a
prioritized task for the developers.

 Presuming processing one order
 would be as simply as one plain instruction, the join order optimizer
 would need on a Pentium III 500 (225.4 MIPS):

 #tables
  10  0.045 secs
  20  0.0047 secs
  30  4.76 secs
  40  1.35 hours
  50 57.81 days
  60162.20 years

 Of course, real scaling will differ as pure MIPS don't show real
 speed. But you get the idea.

Yupp. Exponential growth.

 For your purpose, you have to try out, if speed decrease with 30
 tables matters to you. Please post any result you find.

Yes, I will do some tests during the next couple of weeks. I am currently
working on the import routines, I should be able to test some queries 

RE: Maximum JOINED tables

2002-06-06 Thread Roger Baklund

* Harald Fuchs
 In article [EMAIL PROTECTED],
 Roger Baklund [EMAIL PROTECTED] writes:

  Yes... that is strange. There could be a bug in the crashme
  report, or the limit may have been lowered after 3.23.39, but
  neither of those seems very likely (to me, anyway)... anyone
  got a clue?

 I just tried a multiple self join on 4.0.1 and got the following

 DBD::mysql::st execute failed: Too many tables. MySQL can only
 use 31 tables in a join at ./x2.pl line 47.

Me too, running 3.23.30-gamma on w2k:

mysql desc select * from
- tab1 t1,tab1 t2,tab1 t3,tab1 t4...
ERROR 1116: Too many tables. MySQL can only use 31 tables in a join

So... is the crashme report wrong? Or was the limit really 32 in 3.22.39,
got lowered to 31 in later versions (like 3.23.30-gamma), raised to 63 in
version 3.23.39, and then lowered to 31 again (at least in versions 3.23.47
and 4.0.1)...?

URL: http://www.mysql.com/crashme 

(Select both mysql versions, submit, and search for tables in join on the
result page.)

--
Roger


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: RE: Maximum JOINED tables

2002-06-06 Thread Victoria Reznichenko

Roger,
Thursday, June 06, 2002, 12:08:13 PM, you wrote:

 I just tried a multiple self join on 4.0.1 and got the following

 DBD::mysql::st execute failed: Too many tables. MySQL can only
 use 31 tables in a join at ./x2.pl line 47.

RB Me too, running 3.23.30-gamma on w2k:

mysql desc select * from
RB - tab1 t1,tab1 t2,tab1 t3,tab1 t4...
RB ERROR 1116: Too many tables. MySQL can only use 31 tables in a join

RB So... is the crashme report wrong? Or was the limit really 32 in 3.22.39,
RB got lowered to 31 in later versions (like 3.23.30-gamma), raised to 63 in
RB version 3.23.39, and then lowered to 31 again (at least in versions 3.23.47
RB and 4.0.1)...?

RB URL: http://www.mysql.com/crashme 

RB (Select both mysql versions, submit, and search for tables in join on the
RB result page.)

If you have 32 bit CPU you can use 31 tables in join. 63 tables with
64 bit CPU.

RB --
RB Roger




-- 
For technical support contracts, goto https://order.mysql.com/?ref=ensita
This email is sponsored by Ensita.net http://www.ensita.net/
   __  ___ ___   __
  /  |/  /_ __/ __/ __ \/ /Victoria Reznichenko
 / /|_/ / // /\ \/ /_/ / /__   [EMAIL PROTECTED]
/_/  /_/\_, /___/\___\_\___/   MySQL AB / Ensita.net
   ___/   www.mysql.com




-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-06 Thread Roger Baklund

* Victoria Reznichenko
[...]
 RB So... is the crashme report wrong? Or was the limit really 32
 RB in 3.22.39, got lowered to 31 in later versions (like 3.23.30-
 RB gamma), raised to 63 in version 3.23.39, and then lowered to
 RB 31 again (at least in versions 3.23.47 and 4.0.1)...?

 RB URL: http://www.mysql.com/crashme 

 RB (Select both mysql versions, submit, and search for tables
 RB in join on the result page.)

 If you have 32 bit CPU you can use 31 tables in join. 63 tables with
 64 bit CPU.

ok... thanks.  :)

But... is this the only feature of mysql which is dependant on the processor
architecture?

I'm kind of nervous about this, because I'm currently working on a 'very
normalized' database schema, and it is possible I will need more than 31
tables in a single select... including multiple joins to the same tables, of
course. It would be very bad if my design turned out to be 64-bit
dependant...

--
Roger


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-06 Thread Benjamin Pflugmann

Hi.

On Thu 2002-06-06 at 19:17:10 +0200, [EMAIL PROTECTED] wrote:
[...]

 But... is this the only feature of mysql which is dependant on the
 processor architecture?

See it the other way: You are not garantueed more than 31 tables in a
join. If you happen to have a 64Bit architecture, you can use more.

 I'm kind of nervous about this, because I'm currently working on a 'very
 normalized' database schema, and it is possible I will need more than 31
 tables in a single select... including multiple joins to the same tables, of
 course. It would be very bad if my design turned out to be 64-bit
 dependant...

AFAIK, if you need so many tables, you get a major problem: MySQL
figures out the best join order by trying all combinations. This
becomes increasingly slow as you are going to join more tables. In
other words: Trying to use 30 tables in one select probably will be
too slow as that it would matter whether so many tables are allowed in
a join or not.

Of course, you could work around it by using STRAIGHT_JOIN, but I
assume it is to cumbersome to figure out the optimal join order for
all SELECTs.

Bye,

Benjamin.

-- 
[EMAIL PROTECTED]

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




RE: Maximum JOINED tables

2002-06-05 Thread Weaver, Walt

32.

-Original Message-
From: Arthur Fuller [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, June 05, 2002 9:48 AM
To: mysql mailing list
Subject: Maximum JOINED tables


I have searched the manual without success. I want to know the maximum
number of tables MySQL permits to be JOINed in a single SELECT statement.

TIA,
Arthur


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




RE: Maximum JOINED tables

2002-06-05 Thread Andrew Hazen

Would have been a little spooky if were '42'! :)

Andrew Hazen

-Original Message-
From: Weaver, Walt [mailto:[EMAIL PROTECTED]] 
Sent: Wednesday, June 05, 2002 11:51 AM
To: 'Arthur Fuller'; mysql mailing list
Subject: RE: Maximum JOINED tables

32.

-Original Message-
From: Arthur Fuller [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, June 05, 2002 9:48 AM
To: mysql mailing list
Subject: Maximum JOINED tables


I have searched the manual without success. I want to know the maximum
number of tables MySQL permits to be JOINed in a single SELECT
statement.

TIA,
Arthur


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




RE: Maximum JOINED tables

2002-06-05 Thread Weaver, Walt

Yup, it'd be the answer to everything.   :)

-Original Message-
From: Andrew Hazen [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, June 05, 2002 10:00 AM
To: Weaver, Walt; 'Arthur Fuller'; 'mysql mailing list'
Subject: RE: Maximum JOINED tables


Would have been a little spooky if were '42'! :)

Andrew Hazen

-Original Message-
From: Weaver, Walt [mailto:[EMAIL PROTECTED]] 
Sent: Wednesday, June 05, 2002 11:51 AM
To: 'Arthur Fuller'; mysql mailing list
Subject: RE: Maximum JOINED tables

32.

-Original Message-
From: Arthur Fuller [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, June 05, 2002 9:48 AM
To: mysql mailing list
Subject: Maximum JOINED tables


I have searched the manual without success. I want to know the maximum
number of tables MySQL permits to be JOINed in a single SELECT
statement.

TIA,
Arthur


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-05 Thread Roger Baklund

* Arthur Fuller
 I have searched the manual without success. I want to know the maximum
 number of tables MySQL permits to be JOINed in a single SELECT statement.

According to crashme, it is 32 in 3.22.39 and 63 in 3.23.39.

URL: http://www.mysql.com/information/crash-me.php 

-- 
Roger

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




RE: Maximum JOINED tables

2002-06-05 Thread Weaver, Walt

Hmmm... we're running 3.23.47 here and have run into a 32 table join
limit...

--Walt

-Original Message-
From: Roger Baklund [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, June 05, 2002 10:19 AM
To: mysql mailing list
Cc: Arthur Fuller
Subject: Re: Maximum JOINED tables


* Arthur Fuller
 I have searched the manual without success. I want to know the maximum
 number of tables MySQL permits to be JOINed in a single SELECT statement.

According to crashme, it is 32 in 3.22.39 and 63 in 3.23.39.

URL: http://www.mysql.com/information/crash-me.php 

-- 
Roger

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail
[EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-05 Thread Sabine Richter

And how long does it take to compile a 32-tables-statement?
With the gupta server a 14-tables-statement took 2 hours.

Sabine

Weaver, Walt wrote:
 
 Hmmm... we're running 3.23.47 here and have run into a 32 table join
 limit...
 
 --Walt
 
 -Original Message-
 From: Roger Baklund [mailto:[EMAIL PROTECTED]]
 Sent: Wednesday, June 05, 2002 10:19 AM
 To: mysql mailing list
 Cc: Arthur Fuller
 Subject: Re: Maximum JOINED tables
 
 * Arthur Fuller
  I have searched the manual without success. I want to know the maximum
  number of tables MySQL permits to be JOINed in a single SELECT statement.
 
 According to crashme, it is 32 in 3.22.39 and 63 in 3.23.39.
 
 URL: http://www.mysql.com/information/crash-me.php 
 
 --
 Roger
 
 -
 Before posting, please check:
http://www.mysql.com/manual.php   (the manual)
http://lists.mysql.com/   (the list archive)
 
 To request this thread, e-mail [EMAIL PROTECTED]
 To unsubscribe, e-mail
 [EMAIL PROTECTED]
 Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
 
 -
 Before posting, please check:
http://www.mysql.com/manual.php   (the manual)
http://lists.mysql.com/   (the list archive)
 
 To request this thread, e-mail [EMAIL PROTECTED]
 To unsubscribe, e-mail [EMAIL PROTECTED]
 Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-05 Thread Dan Nelson

In the last episode (Jun 05), Sabine Richter said:
 Weaver, Walt wrote:
  From: Roger Baklund [mailto:[EMAIL PROTECTED]]
  * Arthur Fuller
   I have searched the manual without success. I want to know the
   maximum number of tables MySQL permits to be JOINed in a single
   SELECT statement.
  
  According to crashme, it is 32 in 3.22.39 and 63 in 3.23.39.
  
  Hmmm... we're running 3.23.47 here and have run into a 32 table
  join limit...

 And how long does it take to compile a 32-tables-statement? With the
 gupta server a 14-tables-statement took 2 hours.

That all depends on how complex your query is and the size of the
tables.  A query that takes 2 hours to run probably can be optimized a
lot, though.

-- 
Dan Nelson
[EMAIL PROTECTED]

-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php




Re: Maximum JOINED tables

2002-06-05 Thread Roger Baklund

* Arthur Fuller
  I have searched the manual without success. I want to know the maximum
  number of tables MySQL permits to be JOINed in a single SELECT
  statement.

* Roger Baklund
 According to crashme, it is 32 in 3.22.39 and 63 in 3.23.39.

 URL: http://www.mysql.com/information/crash-me.php 

* Weaver, Walt
 Hmmm... we're running 3.23.47 here and have run into a 32 table join
 limit...

Yes... that is strange. There could be a bug in the crashme report, or the
limit may have been lowered after 3.23.39, but neither of those seems very
likely (to me, anyway)... anyone got a clue?

--
Roger


-
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/   (the list archive)

To request this thread, e-mail [EMAIL PROTECTED]
To unsubscribe, e-mail [EMAIL PROTECTED]
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php