About LaurieF

LaurieF · ‎08-06-2019

Want to add on something really stupid? This is from the days before %sysfunc, putn and the z format... %let j = %substr(000&j, %length(&j), 4); For values of j up to 9999, right-justify and zero-fill j. Want longer numbers? Add zeros, and change the last parameter. I used this construct for years, but the documentation was usually fairly thorough.

LaurieF · ‎07-23-2019

If you know for sure that at least one row will have a populated pounds value, put the following as a statement in your proc import: guessingrows=n; where n is the number of rows you want to scan before you do the import. Once the import process has detected that, "Oh yes, I see now. This pound column is numeric! Silly me!", it'll change all the blank cells to numeric missing. And your subsequent code should work. It saves programmatic work-arounds.

LaurieF · ‎07-18-2019

If there's no index on the column, yes: it'll do a table scan. If you've got an index on it though, it should be quite responsive. I'd be surprised if it'd take more than a few seconds then.

LaurieF · ‎07-17-2019

When I create a transformation, the only code I put in Source Code is call to the macro which executes the code. That way whenever the code has gone through unit-testing, all the jobs which use that transformation don't have to be re-deployed. However when the transformation acquires new options, as one of mine did yesterday, my usual approach has been to: sigh analyse the transformation to check the jobs which use it redeploy the jobs But now I have dozens of jobs that use the transformation, and I couldn't be bothered sighing. I knew that the majority of the jobs wouldn't need to modified, and would get by on the defaults; however the code needed to be aware of the newly created option names and not fall over if the job didn't create them. The fix is really quite easy, thanks to %symexist. Does the macro exist? No? Then create it, and set it to the same default as the updated parameter requires. %if %eval(not %symexist(history_type)) %then %do; %global history_type; %let history_type = %nrquote(H2); /* Default, standard processing */ %end; The new jobs, with history_type defined via the updated transformation, simply bypass the logic. The old jobs, without it, create history_type and populate it. I appreciate that it won't always be as simple as this, but it saved me a lot of time. And meant that I could go off for my lunch early.

LaurieF · ‎07-17-2019

The line order datepart(c.lastup_DT) Between '17JUL2018'd and '17JUL2019'd; forces the rows between those dates to the bottom of the output - don't you mean where instead of order? Assuming that you do, this would be better: where c.lastup_DT Between '17JUL2018:0:0'dt and '17JUL2019:23:59:59'dt; That way any null values won't trigger, and there won't be any warning messages.

LaurieF · ‎07-14-2019

I generally concur, but not completely. And I say this as a big fan of compression. If the PDV is predominantly numeric (especially with few missing values), or otherwise contains short fully populated character variables, compression can well increase the size of the dataset. The extra I/Os required to read the larger-than-necessary dataset, with the extra cost of decompressing, can thus extend the run-time considerably. My rule of thumb is: if compression is less than 30%, don't bother.

LaurieF · ‎07-09-2019

I was asserting that it wasn't anything to worry about. From memory, they're stored in the length assigned, but only take up three bytes. Any length longer will just pad the extra with '00'x. That's in Windows - other operating systems will do it differently - big- and little-endian, and that sort of thing. SAS wouldn't dare change it, because all the code in the world would break!

LaurieF · ‎07-09-2019

3 is the minimum. Unless you're on an IBM mainframe, when I think it's 2!

LaurieF · ‎07-04-2019

And it works - well spotted. I owe you a beer. Now, down to examining the stuff about each job. Seats upright, table stowed, ready to extend this metaphor too far...

LaurieF · ‎07-04-2019

Really? I'm an idiot! Thanks very much. I can't wait to get to work tomorrow morning to try it out.

LaurieF · ‎06-30-2019

This may look like over-egging the pudding, but it allows for any variations in the population of your source data. #ThinkingAhead data start; infile cards dsd dlm=',' firstobs=2; attrib id length=3; attrib test length=$ 1; input id test; cards; id,test 1,A 1,B 2,B 3,A 3,B 3,B 4,A 4,B 5,A ; run; proc sort data=start noequals; by id; run; data target; set start; by id; attrib test_a test_b length=$ 1; retain test_a test_b; if first.id then call missing(test_a, test_b); select(test); when('A') test_a = 'Y'; when('B') test_b = 'Y'; otherwise; end; if last.id; if test_a ne ' ' and test_b ne ' '; keep id; run;

LaurieF · ‎06-30-2019

Sadly I haven't been able to get Global Hosting to switch it on for me...

LaurieF · ‎06-30-2019

Here you go: data _null_; length id folder_uri job_uri member_uri uri $ 20 type prop value $ 256; length name _uri location _location $200; call missing(name, _uri, location, _location, uri); call missing(id, type, folder_uri, job_uri, member_uri, prop, value); rc = metadata_pathobj(' ', '/DIP Data Management/5000 Processes/10 Raw Ingestion', 'Folder', type, _uri); do n = 1 by 1 while(metadata_getnasn(_uri, 'Members', n, member_uri) > 0); put n= _uri= member_uri=; end; run; The output I get shows that it's nearly there - but not quite. I know the number of members, but the uri returned from getnasn is the folder's one. n=1 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=2 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=3 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=4 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=5 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=6 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=7 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=8 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=9 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=10 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=11 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=12 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=13 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=14 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. n=15 _uri=A5IKCEC7.AA00008A member_uri=OMSOBJ:Job\A5IKCEC7. etc...

LaurieF · ‎06-27-2019

I'm afraid you've missed the point. I was talking about how to read members in a tree. The other problem, which I've already solved, was to show that I do know how to navigate my way through metadata functions. In that particular case we have 20-30 entries, and each one previously had to be updated every time we did a migration - the import function did not update them to the correct values. This is beyond a manual process and has no part in a migration document.

LaurieF · ‎06-26-2019

When we move from dev to test (and to prod), our Hadoop and CAS libraries and jobs have to be manually patched because the text strings contained within the library references frustratingly don't remap to the test properties and associations, like it would with a real DBMS. (Old man shouting at clouds) The library bit is relatively easy, because I can read through all the "omsobj: SASLibrary?@Id contains '.'" entries where the engine is CAS, HADOOP or IMPALA, then walk my way though the associated data and set the attributes and properties and patch them. The execution time is virtually instantaneous, and works well. But the jobs (and it's mainly the CAS ones that are causing me grief) are proving problematic. I have three or so metadata folders whose uris I can identify with metadata_pathobj. I can determine the number of members of each tree, using metadata_getnasn with Members as the association. In fact do n = 1 by 1 while(metadata_getnasn(_uri, 'Members', n, member_uri) > 0) loops the expected 20 times (for the particular folder I'm initially interested in. But nothing shows up in member_uri. What am I missing? Am I using the wrong association? How do I walk through a tree? (If you'll excuse the expression)

Online Status	Offline
Date Last Visited	‎10-28-2024 09:00 PM