DSXchange

256 is the maximum number of columns that can be returned in a result set from MS Access. There is no limit in hashed files. Obviously, though, if you push the boundaries, you're defeating the aim of efficiency; whether it's a hashed file or a database table, a SELECT that retrieves two columns is g...

OR... ... because the Transformer stage is doing a left outer join, you could switch streams (Department becomes your main stream), use the multi-row lookup capability of the ODBC stage and lie that DEPTNO is the key to the EMP table (to force the parameter marker into the WHERE clause). Note that y...

Point 1 - there's no such thing as a "lookup stage" in server jobs. However, I believe that's just a terminological glitch on your part, as you've called it a Transformer stage in your subject. Point 2 - a reference lookup in a Transformer stage is effectively performing a left outer join, in that y...

Your approach relies on routine arguments continuing to be passed by reference rather than by value. There's no such guarantee, in fact this is why warnings are generated by the Routine pre-compiler when it detects any attempt to change the value of any argument. @USER0 through @USER4, on the other ...

Are you referring to a Command Activity, which is used in a job sequence? As far as I know there's no Command Stage per se for server jobs, but it is really easy to create a transform function that executes a command based on the arguments passed to it. Then you have to figure out what to pass back,...

That information is already logged for you, in an "active stage finishing" message. Why do you want to do it separately? The answer to your question is that you do need to do it in a routine. If you try to do it, say, in a column derivation, then it occurs every row, which will slow your job down hu...

An alternative is to update five user variables @USER0 through @USER4 in the routine. These are available in the stage that invokes the routine. For example, I have a routine that parses a telephone number into country code, area code, carrier id and line number and returns these through @USER1 thro...

I think it's time for you to post the exact: (a) column definitions from the main input link (b) column definitions from the reference input link (c) reference key expression (d) any constraint expression (e) samples of data from main input link with text fields quoted (f) samples of data from refer...

Should we give up on the whole idea of having a single large hash table available to us and just create say hlp_dim_keys_hash_001/002/003 etc so we break the hash table up into smaller parts and let DS deal with these smaller parts? If memory caching is important to you, that's what I'd advise. Alt...

It's a decision you have to take. You can either create DataStage jobs to generate text files that the scripts can use to load into the production tables, or you can design DataStage jobs that replace the functionality of the scripts. Either is possible, even a hybrid approach where the logic of the...

It IS completely different. Server jobs are build on DataStage BASIC, parallel jobs employ "Orchestrate shell" and custom coding is done in C. In version 7 you do at least get the "BASIC Transfomer" stage, which allows you to re-use the derivation and constraint expressions you may already have deve...

Messages about "harness" usually relate to the "on the fly" test program that is written to test your routine when you click the "Run" or "Run All" button.
Are you running out of disk space, or have unusual permissions, in the project directory, particularly the DSU_BP and DSU_BP.O directories?

I'd just like to reinforce as strongly as possible everything Vincent posted. He was even nice enough to use US spelling for you! In particular, if you are going to "roll your own" do not stint on planning and design time. US Postal certification is available to plug in directly to Quality Stage (wh...

I think the code you need to populate the hashed file should be and that the hashed file's key column should be web_text.
Is this the case?

Nah. Kim's not correct in stating an upper limit of 4GB for 64-bit addressing in hashed files. 64-bit addressing allows a theoretical upper limit of 19 million TB for a hashed file. Some operating systems, however, restrict (!) you to one million TB. Thomas, RESIZE filename * * * 64BIT [ USING pathn...