DSXchange

rohitash_saini wrote:Can you please share this spreadsheet ?

Done, link updated in the first post of the main thread. Dropbox changed the way public shared files worked, it's now on Google Drive. If anyone has a better place to host it, please feel free to "mirror" it and I will include a link in my post.

The code in the OP contains a memory leak. It is allocating memory with the new operator, and that memory never gets freed. How to fix this? Sorry, not possible. At present it is impossible to write string functions that do not have memory leaks or thread safety issues. IBM should implement a parall...

pxEreplace is a bad idea. It is impossible to implement a string replacement function in DataStage without introducing either contention or a memory leak*. Sorry. What IBM desperately need to do is to allow std::string parameters (and return type) for parallel routines. That is, after all, precisely...

If you look at my second post, you can see that the column orders are the same. I can't remember what the solution to this was since it was a long time ago but I think I basically re-built the stage from scratch and it worked, and I never had the time to actually find out what the difference was.

If someone's going to update the animated GIF then perhaps they could complete the retro '90s web experience and add some flying toasters and maybe a dancing baby as well.

According to whois, their web host is IXWEBHOSTING.COM so maybe you should contact them. Their IP address is owned by United States Columbus Ecommerce Corporation (ecommerce.com) so maybe abuse@ecommerce.com is an option.

saidvs wrote:try this
Convert('KolkataMumai','Calcutta','Kolkata') it works in parallel transformer too..

That will convert the third parameter replacing 'K' -> 'C', 'o's -> 'a', 'k' -> 'c', 'a' -> 'u', etc., and removing all 'M', 'u', 'm' and 'i' characters. Not exactly what the OP wanted.

The problem is, you can't pass in a char*. DataStage can pass in a char* but I, as a DataStage user, can only pass in a Stage Variable or a string literal. DataStage will somehow construct a char array and pass a pointer to that, but as far as I know it is unsafe for the function to write to that ar...

That's nice, but I'd still like to be able to implement the pxStrFilter routine that is in the third post of this thread.

IBM have finally given a sensible response. Good Morning, I did some further research and found some more information: It would seem that there is no way to free the memory that is allocated in a C transform routine. You should not use malloc in a transform routine. We suggest that if you have a com...

Sadly I must advise that you DO NOT use the pxEreplace function. This is from IBM support: Good Morning, I did some further research and found some more information: It would seem that there is no way to free the memory that is allocated in a C transform routine. You should not use malloc in a trans...

That's pretty much the same as killing the Designer task, since Designer will have to be closed after you do that, and I think it's also necessary (under v7 anyway) to clean up the dsapi_slave process after killing Designer anyway.

Target result should be col1 col2 col3 105 500 506 105 101 201 Why should this be the output? What's the rule that rejects the second row? At a guess, I'd be tempted to say "where col2<>b_col2 or (col2=b_col2 and col3=b_col3)", is that it? In english, "if Col2 matches then Col3 has t...

If you're going with a Transformer based approach, remember to include "Or LastRow()" in the output constraint in case your data doesn't have an exact multiple of three rows. I'm not sure I like the idea of concatenating with spaces and then splitting. I'd rather gather the values in three...

I wrote a Parallel Routine tro do a similar task - search for pxStrFirstCharList for the version that finds the first occurrence of a character in a set of characters, it should be easy to create a version that finds the last, I did make one but I don't have the source for that to hand.