SQL Power Software Forum

what happen if you just ignore the error message? are the connections still there next time you run the pl? if they are, you should be just OK.

maybe you have too many connections defined in your power loader, try clean up something you don't need first.

call your sql script from a batch script, that should do it.

SQL Anywhere is not fully supported by the engine, for example, we I have not finish some functions that needed for sequence generating etc., but you can always extract data from it.

what kind of output you are looking for? the engine should be able to capture the output of SQL*PLUS if you run a script in a job, and I think it should be able to get the dbms.outline of package, but maybe be only true for stored procedure in a job, and maybe you can try debug mode.

substring is a OK name for the engine, I guess you might have other problems using the function, try to run it in debug mode, exam the input value of the function, if the input value is a string, it should be quoted by ', if that is a number, then quoting is not necessary.

There is no limit on the level of nested jobs, but there are limits on other things, like number of total connection, number of output on a interface, and the database may also has limit of open cursor for a connection, so if you have a big job that read and write from/to many databases, try to debug run it first!

No, the variable will be reassigned.

I guess you have this question because you are confused of the mapping sequence, the key is the direct mapping. the engine try to fill a output column by go through the mapping steps, for the first time that the variable appear on the mapping, it's null by default, then, the engine can get value from translation, group function, sequence, default etc. but direct mapping, then assign value back to the variable, next time, when the variable appear on the mapping again, any new value happen before direct mapping will be assign back to the variable, otherwise, the value stored in the variable will be used in the direct mapping. if that is not null, the rest of the mapping steps will be ignored.

That's right, any column could relate to any column in a database, but all we need is define and save a simple relationship such as table.columns -> table.columns in somewhere, like table or project preference.
for example, in the project that I am currently working on, we use a lot of 'code tables', the code tables store the code to description mapping, then use the code in the real tables, a few related (or maybe not) real tables share one code table, using another column in the code table as type to catagorize the codes. what I mean is there are no real database relationship between real tables and the code tables, but the business or the design defined such relationship. it's just a part of the project/design.

since Jonathan pointed out that view table content feature is in the develop plan in another post:
http://www.sqlpower.ca/forum/posts/list/1718.page

here are some thought I have about view table content for a while, I have been keep thinking of it because it's painful when you need this feature but none of the database tool has it as far as I know, I will try to explain that as clear as possible.
Here is what I need and I think it's useful for many of us:
in many time, we need to query a table, to get some data (i.e. id,name,date etc.) then use that to query some other related tables to get detail data. in that case, I have to open another table from the object browser, copy and paste the data to filter/statement etc. (take a database tool for example, TOA* or dbvisual***, you have to do almost the same thing for each related table), then you go back to the master table, you may or may not have the filter any more.

here is my suggestion of the new feature:
open up a table, use viewer content filter or database filter, get a smaller result set, show the related tables(child table or parent table, does not matter, even the tables with virtual relationship) around it, select one or more column, use that as master record to find related records in the related tables, we may want to support a few match methods like exact match, alike, range etc. and we may want to use color or something to show child found or not found of the master records.

the feature is very useful when you need to explain the legacy system, or debug the result of your ETL interfaces. that all for now, let me know what do you think.

there is another setting you can try, since you are working on the meta-data, update PRIMARY_KEY_INDEX to null, put the pk column names in INDEX_COLUMN_NAME[0-9], if your update duplicate rows does not cause pk violation, the engine may not delete the duplicate rows (according to a quick view of the code

.

did the engine tried to delete duplicate rows in all merge tables?
well, there is another logic in the engine, if the it's merging the last table (last table in the merge list, which should be the master table, yes, this table should be the first and the last, if I remember that right), and we have the index name in the meta-data, then the duplicate rows will be deleted.

this behavior is controlled by meta data column pl_merge_criteria.DELETE_DUP_IND and another rule, try change it by using the front-end. the other rule is: if the update to the table causes pk duplicate, then then dup must be deleted.

Well, I am glad you find out the work around, after these years use of the engine, we never have this problem as far as I know, number as data type in a table is common in oracle database, but we did not have the problem, I wonder how did this happen, what did you do to your database? what kind of local did you have? because the engine does not makeup the data type itself, the only reason that the engine think the column is double is because in the meta-data of the column(return by oracle), the column is labeled as 'double' or 'double precision', which we have never seen in oracle before.

Was there any problem when create the MATCHMAKER_TEST00 table?