DSXchange

Well, I know that too broad of a search overwhelms the indexes and returns nothing. You can imagine that the word "job" has about 10 million uses in over 69000 posts, so, it is a bad choice to use when doing an any-word match.

Try using the exact error message, or a specific term.

No covert book writing going on here...no sir-ee...just minding my own business...we're all just friends, right?...I'll be just moving along...

As your lookups are busy with the round trip between DS and Oracle, your DS processes get to take a nap. Because of round trip latency, query queing, and query contention (using IPC stages to get 6 transformers independently executing with one lookup each) because 6 queries are firing for each row, ...

Are you sure DTE_INT_BEGIN is a column in the inbound link?

By compatible, you mean using the Oracle OCI stages? The answer is complicated, becuase the OCI client libraries supplied with Oracle work with specific databases. DS has OCI8 and OCI9, so if you're using OCI9, the Oracle 9i client works with 10g database. The OCI8 with Oracle 8i client works with 1...

try ip 127.0.0.1 or 'localhost' from the DS Server. If the engine is running, this WILL work.

Okay, you can set a variable to be the result of a word count for lines on a data file. Then, subtract 1 from that value, that's the result to use with head. head gives you the top N lines in a file, thus original file line count minus 1 would stop at the trailer record. Pipe that to tail with the o...

rgandra wrote:If you know the format of the header and trailer record, you cna filter those records by adding contraint in the Transformer stage.

No, PX is very sensitive to metadata inconsistencies in the Sequential file stage.

Are you talking about maintaining foreign key relationships? Meaning a parent and children records are processing today, and if any of the related rows fail to load, all need to fail? If you have a parent fail during processing, you can easily put those parents into an "avoid" lookup so that childre...

Do you still have a question?

I don't think we have enough information, hence the lack the of responses. So we have this start date and a duration. Sure we can figure out the number of weeks and then compute the Sundays that fall in that range. What records are you talking about? Do you just need a function that returns an array...

What does matter is the scale. When there is a scale setting and you perform math (just a "+0" will do) on the column, the number is rounded to to the setting of the scale if the data type is DECIMAL. Other than that, there's no interaction with the metadata during processing until you reach a datab...

Keep it small, less than 30 characters, because the dialog boxes are small. If you have too big of a name, you can't distinguish project_xyz_1 from projecT_xyz_1234 because of right side truncation.

Read up on the hash files in the most recent newsletter. There are no datatypes, no practical size limitations, on columns.

kumar_s wrote:I wonder if IBM will have a look on this thread

Someone there reads every post...