Question

I have the following code, and database. My intent is the end user can select the colums from a database (by ‘ajax’ing option/selects direct from mysql) and run a report on them. I can predetermine the tables they use, they pick the columns.
my biggest problem is the BEST way to perform a ‘schedule’ query.

To run a schedule i need the Details from ‘Customer’ and ‘Delivery’ – (customer info and delivery address) and i use the ‘custnum’ from all tables, plus the year, to choose which yearly schedule to run – ie custnum bar12345 – yearly schedule table is bar123452012

there could be 2 products per ‘custnum’ pulled from the bar123452012 table.

this table is just a single row per product, but 367 columns, 1- prodid, and 366 titled with the
yday
integer. php works out if it needs to use 365 or 366 in that year in a different function.

What would you suggest? A php loop to pull all customers one by one and format in a table, or is there an SQL query that will Left Join a different table to each row based on the tindex . 2012id?!
need speed and efficiency as we are talking at least 200 customer tables, maybe up to 1000.

Thanks to Dwurf for a reality check. Reorganised database to have one schedule table, and work from that.

Thanks Dwurf. New design being worked on all afternoon… Database

Table structure for table customer

custnum varchar(255) No
surname varchar(50) No
name varchar(50) No
phone1 varchar(50) No
phone2 varchar(50) No
email varchar(50) No
add1 varchar(50) No
add2 varchar(50) No
city varchar(50) No
postcode varchar(10) No

Table structure for table deliverydetails

special varchar(50) No
add1 varchar(50) No
add2 varchar(50) No
city varchar(50) No
postcode varchar(10) No
custnum varchar(100) No

Table structure for table notes

custnum varchar(50) No
notes varchar(50) No
time timestamp No CURRENT_TIMESTAMP

Table structure for table prodgroups

pgroup varchar(20) No
name varchar(50) No
description varchar(50) No

Table structure for table products

name varchar(50) No
quantity varchar(50) No
description varchar(50) No
pgroup varchar(50) No
price varchar(50) No
picture varchar(100) No
productid varchar(10) No
suppid varchar(50) No
vat tinyint(1) No

Table structure for table schedule

productid varchar(15) No

d0 varchar(100) No
d1 varchar(100) Yes NULL
d2 varchar(100) Yes NULL
…….
d362 varchar(100) Yes NULL
d363 varchar(100) Yes NULL
d364 varchar(100) Yes NULL
d365 varchar(100) Yes NULL

custnum varchar(50) No

id int(50) No

Table structure for table suppliers

suppid varchar(50) No
name varchar(50) No
phone1 int(100) No
phone2 int(100) No
email varchar(100) No
add1 varchar(50) No
add2 varchar(50) No
city varchar(50) No
postcode varchar(10) No
description varchar(100) No

Table structure for table vatgroups

vat int(20) No
amount decimal(20,0) No
description varchar(20) No

Answer 1

What would you suggest? A php loop to pull all customers one by one and format in a table,

This approach is suicidal for performance

or is there an SQL query that will Left Join a different table to each row based on the tindex . 2012id?!

This isn’t really possible with the design you outline in your question. A design change might help though: if you have a single table for all of your reports and you add a column for the report name you will be a bit closer. Consider something like:

Table structure for table Reports
custnum varchar(255) No
productid   varchar(15) No  
date  DateTime    No  

Now you can construct a query to grab all rows for a particular customer

select * from Reports 
where date between '2012-01-01' and '2013-01-01'
and custnum = 'bob54326'

Use PHP (or whatever you’re using) to generate the table from these results.

I know this strategy breaks your ‘direct table access from UI’ model, but I’m afraid that model simply doesn’t work very well. Change your design to work with the tools you have instead of trying to force the tools to conform to your design.