DSXchange

battaliou

Cool, post the code. I would have passed the invocation id down to the sub job as a seperate parameter...

battaliou

If you have to use SQL to extract table usage from your repository, maybe you should consider setting up a few i desciptors? You'll have to issue the following commands from Administrator: ED DICT DS_DSJOBOBJECTS ID1 I 0001 = I 0002 = @RECORD<13,7> 0003 = ----: FI ED DICT DS_DSJOBOBJECTS ID2 I 0001 ...

battaliou

Hello thiagol If you've established your number of columns, you could always "doctor" your incoming data in a transformer which adds the extra columns onto your source. To do this, you simply read your input data as a single string (ie only one column in the source stage), use dcount to es...

battaliou

I can confirm that this is the case. I got hold of the NotesSQL301.exe file, uninstalled 3.02, rebooted, installed 3.01 and now my odbc drivers work as expected.

battaliou

Inside the datastage administrator, in any project, execute the following command:

Code: Select all

CT VOC RELLEVEL

This will give you your release level.

battaliou

Hello Arnd

Converting to NLS was very easy. The knock on effect only seemed to affect dates. For example a standard date in non nls looks something like 2005/02/25 whereas in NLS its 2005 02 25.

Code: Select all

OCONV(DATE(),'DYMD[4,2,2]')

This issue affected some of our function calls.

Regards
Martin Battaliou

battaliou

I edited my uvconfig file (NLSMODE 0) to disable NLS in my current environment. I had to do a uvregen as well, but everything reverted ok.

battaliou

I got a very similar error from DB2 on an AS400:

Package "NULLID .SYSSH000" was not found. SQLSTATE=51002 ... DSP.Open GCI $DSP.Open error -100.

Solution was to bind the database from the ODBC settings.

battaliou

Your job will run ok, but this is not best practice in EE as you should specifically define type conversions. I think you will find a slight gain in performance on your job too.

battaliou

These questions look like they've been drawn up by people who have little experience in coding the product. I'm particularly appalled at the "read only routine" and universe configuration questions. The answer here surely is "Why?" If you are concerned with any of the above issue...

battaliou

You could use BASIC:

Code: Select all

openseq 'in.txt' to f else return
openseq 'out.txt' to t else return
loop
 out = ''
 for i = 1 to 3
  readseq rec from f else return
  if out = '' then out = rec else out := ',':field(rec,' ',2)
 next i
 writeseq out to t else return
until 1=2 do repeat

battaliou

I'm still keen on the SQL relational approach. One of the problems I have using hash files is that the source data seems to have a one to many relationship e.g. column matnr may not be unique in file mvke.txt. An outer join will fix that in Oracle by returning all of the relationships. Maybe you cou...

battaliou

I think your question regards sorting data in parallel streams? If this is the case, better to partition your data along your sort critria, and sort seperately. You'll find an exponential improvement in performance. Of course there is a memory impact on using multiple sorting (especially if your par...

battaliou

I dont want to cause any derision here, but the code posted above will need to be modified as the datastage logger will have problems accessing the file variable FileVar (improper data type) You would probably be better off creating a new variable e.g. FileName = Arg2:'\':Arg1 and using that instead...

battaliou

Osh code is generated by the client, ie Designer, so no unix commands can help you to compile here. Remember you should use "force compile" to ensure the entire job is compiled. Why is your "reset job" not working? Simply generating the same osh should have no benefit for you. I ...

DSXchange

Search found 153 matches

EE sort