I need help with selecting from 2 identical tables in 2 separate databases

[phillip.s.powell]

MySQL 3.23.58 - 4.0.17 (yep, several database server instances, don't
ask)

I have database Spring with table Students

I have database Summer with table Students

I am tasked to produce a query of all students in both tables with no
duplicates. No clue whatsoever.

I'm thinking an INTERSECT followed by 2 UNION SELECT statements, but
how far off am I? The tables are absolutely identical in every way
including indexes; the data MIGHT be identical (same data might be in
Spring.Students and Summer.Students ), however, that won't always be the
case. The column Spring.Students .id will not share the same value as
Summer.Students .id as data can be entered within separate sequences
even if the data in both tables is absolutely identical.

Am I on the right track with INTERSECT/UNION or what do you recommend?

Thanx
Phil

[Bill Karwin]

ph************* *@gmail.com wrote:

MySQL 3.23.58 - 4.0.17 (yep, several database server instances, don't
ask)

I have database Spring with table Students

I have database Summer with table Students

I am tasked to produce a query of all students in both tables with no
duplicates. No clue whatsoever.

I'm thinking an INTERSECT followed by 2 UNION SELECT statements, but
how far off am I? The tables are absolutely identical in every way
including indexes; the data MIGHT be identical (same data might be in
Spring.Students and Summer.Students ), however, that won't always be the
case. The column Spring.Students .id will not share the same value as
Summer.Students .id as data can be entered within separate sequences
even if the data in both tables is absolutely identical.

Am I on the right track with INTERSECT/UNION or what do you recommend?

Thanx
Phil

Well, this would be easier if you had both tables in one database. I
think it's more straightforward and portable to compare tables that
exist in the same database.

If necessary, create a third database and copy the Student tables from
Spring and Summer into the third database. You can use the mysqldump
tool to create a backup of a single table from each, then edit the two
dump files with a text editor to make sure the table names will be
distinct, then restore both dumps to the third database.

Here's one query to get the set of all students. Get all students who
exist in both Spring and Summer, then all students only in Spring, then
all students only in Summer. These three sets have no overlap.

(SELECT S1.student
FROM Spring AS S1 INNER JOIN Summer AS U1 ON S1.student = U1.student)
UNION
(SELECT S2.student
FROM Spring AS S2 LEFT OUTER JOIN Summer AS U2 ON S2.student = U2.student
WHERE U2.student IS NULL)
UNION
(SELECT U3.student
FROM Spring AS S3 RIGHT OUTER JOIN Summer AS U3 ON S3.student = U3.student
WHERE S3.student IS NULL)
ORDER BY student;

Regards,
Bill K.

[Markus Popp]

The right thing is to use a UNION join:

select * from Spring.Students
union
select * from Summer.Students

Markus

[phillip.s.powell]

Unfortunately that choice produced a MySQL syntax error.. I should have
mentioned we're using MySQL 3.23.58 in the current version and will
hopefully be upgrading to MySQL 4.0+ in the new version next year

Phil

Markus Popp wrote:

The right thing is to use a UNION join:

select * from Spring.Students
union
select * from Summer.Students

Markus

[phillip.s.powell]

See below, thanx

Bill Karwin wrote:

ph************* *@gmail.com wrote:

MySQL 3.23.58 - 4.0.17 (yep, several database server instances, don't
ask)

I have database Spring with table Students

I have database Summer with table Students

I am tasked to produce a query of all students in both tables with no
duplicates. No clue whatsoever.

I'm thinking an INTERSECT followed by 2 UNION SELECT statements, but
how far off am I? The tables are absolutely identical in every way
including indexes; the data MIGHT be identical (same data might be in
Spring.Students and Summer.Students ), however, that won't always be the
case. The column Spring.Students .id will not share the same value as
Summer.Students .id as data can be entered within separate sequences
even if the data in both tables is absolutely identical.

Am I on the right track with INTERSECT/UNION or what do you recommend?

Thanx
Phil

Well, this would be easier if you had both tables in one database. I
think it's more straightforward and portable to compare tables that
exist in the same database.

Sadly, that is not an option. The client requested they be put into
separate databases. The explanation was about as "non-techie" as it
gets so I'll just spare you, but they plan to "rejoin" the tables into
one table in the new version of the application next year
If necessary, create a third database and copy the Student tables from
Spring and Summer into the third database. You can use the mysqldump
tool to create a backup of a single table from each, then edit the two
dump files with a text editor to make sure the table names will be
distinct, then restore both dumps to the third database.

The only problem with this approach is that the separate databases are
being currently used right now in a series of web application "patch"
scripts that were written in the zero-hour to placate the customer's
need to see the students separated by "spring" and "summer", so I
cannot yet rejoin them until the customer blesses the workflow
methodology of the updated web application, which won't be until next
year..
Here's one query to get the set of all students. Get all students who
exist in both Spring and Summer, then all students only in Spring, then
all students only in Summer. These three sets have no overlap.

(SELECT S1.student
FROM Spring AS S1 INNER JOIN Summer AS U1 ON S1.student = U1.student)
UNION
(SELECT S2.student
FROM Spring AS S2 LEFT OUTER JOIN Summer AS U2 ON S2.student = U2.student
WHERE U2.student IS NULL)
UNION
(SELECT U3.student
FROM Spring AS S3 RIGHT OUTER JOIN Summer AS U3 ON S3.student = U3.student
WHERE S3.student IS NULL)
ORDER BY student;

Yeah I tried the UNION statement but it failed due to a MySQL syntax
error. I forgot to mention that we're using MySQL 3.23.58 right now
and upgrading is "not an option at this time" (yep, next year.. blah
blah blah)
Regards,
Bill K.

*sigh* Can't use UNION so don't know what to do at this point

Phil

[phillip.s.powell]

Ok, after re-reading (and getting my second cup of coffee) my brain
just kicked into gear and realized what you just said. In short, Phil
has left <stupid&g t; mode.

That sounds doable, however, I will have to work on the means of
ensuring that the data from Spring.Students doesn't produce dups with
Summer.Students , and because UNION is not supported in MySQL 3.23.58
(I've tried to use it and consistently got syntax errors), I have
absolutely no idea how to do this :(

Phil

Bill Karwin wrote:

If necessary, create a third database and copy the Student tables from
Spring and Summer into the third database. You can use the mysqldump
tool to create a backup of a single table from each, then edit the two
dump files with a text editor to make sure the table names will be
distinct, then restore both dumps to the third database.

[phillip.s.powell]

Incidentally, I tried your query, and, unfortunately, it produces a
syntax error in MySQL 3.23.58, which is really unfortunate considering
it's the easiest way to do this, and, at this point, don't have any
alternatives. :(

Phil
Bill Karwin wrote:

(SELECT S1.student
FROM Spring AS S1 INNER JOIN Summer AS U1 ON S1.student = U1.student)
UNION
(SELECT S2.student
FROM Spring AS S2 LEFT OUTER JOIN Summer AS U2 ON S2.student = U2.student
WHERE U2.student IS NULL)
UNION
(SELECT U3.student
FROM Spring AS S3 RIGHT OUTER JOIN Summer AS U3 ON S3.student = U3.student
WHERE S3.student IS NULL)
ORDER BY student;

Regards,
Bill K.

[Markus Popp]

I'm afraid, it's not possible to do this with MySQL 3.23. You could only
query the tables separately and combine them on the client side.

If you consider updating (in fact, it's certainly a good choice to update),
I would recommand that you update straight to 5.0.x. Of course, there are
many details to take care of, but the manual stresses everything you need to
know to update from each release to another.

Markus

[Markus Popp]

Hold on - there's one thing I forgot, there maybe is a solution.

If you use MyISAM tables you can use MERGE tables to combine 2 tables which
are identical in its structure. I'm not absolutely sure if MERGE tables are
supported in MySQL 3.23 (but I think so) and if it's possible to combine 2
tables that are in different databases. But in general, it works like this:

create table table_name ([column_definiti on as in the tables]) engine=merge
union=(table1, table2, ...)

Then this new table holds all records that are in each of the tables and you
can query them together.

Markus

[phillip.s.powell]

Hmm.. I had no idea about that one, but this worked for me:

CREATE TABLE blah SELECT DISTINCT s.* FROM db1.table1 s LEFT OUTER JOIN
db2.table1 u WHERE lower(s.first_n ame) <> lower(u.first_n ame) AND
lower(s.last_na me) <> lower(u.last_na me)

Phil

Markus Popp wrote:

Hold on - there's one thing I forgot, there maybe is a solution.

If you use MyISAM tables you can use MERGE tables to combine 2 tables which
are identical in its structure. I'm not absolutely sure if MERGE tables are
supported in MySQL 3.23 (but I think so) and if it's possible to combine 2
tables that are in different databases. But in general, it works like this:

create table table_name ([column_definiti on as in the tables]) engine=merge
union=(table1, table2, ...)

Then this new table holds all records that are in each of the tables and you
can query them together.

Markus