Reeza, This is what my code looks like I used the automap option (but I tried the XML Mapper too... maybe there's an option in the XML mapper that I missed) filename  XML_file 'C:\folder1\xml_file_1.xml'; filename  XML_MAP  'C:\folder1\XML_MAP1.map'; libname   XML_file xmlv2 automap=replace xmlmap=XML_MAP access=READONLY; proc sql; create table Test1 as    select    DISTINCT              /* distinct is probably optional */              t1.Header_ORDINAL          as Header_ORDINAL             ,t1.QueryId                 as Header_QueryId                      ,t4.response_ORDINAL        as Response_ORDINAL             ,t1.Status                  as Header_Status                      ,t1.TransactionId           as Header_TransactionId               ,t2.Exceptions_ORDINAL      as Exceptions_ORDINAL                 /* . . . many fields . . . */    from      XML_file.Header        t1    left join XML_file.Exceptions    t2   on t1.Header_ORDINAL      = t2.Header_ORDINAL    left join XML_file.Item          t3   on t2.Exceptions_ORDINAL  = t3.Exceptions_ORDINAL    left join XML_file.response      t4   on t1.response_ORDINAL    = t4.response_ORDINAL     /* . . . many tables . . . */ ;quit; If I look at the contents of all the datasets in the XML_file, I can see which Ordinals go with other Ordinals in other tables... so I can do this by hand... hoping for an automatic option. Also, I don't really know if Excel does it correctly; I just know it does it. ... View more

PhilC, thank you for this suggestion. That thread #245030 talks about vertical stacking -- an interesting and efficient way to stack 100 datasets vertically. Unfortunately, I need merging -- side-by-side merging on key fields -- not vertical stacking ... and my variables are different in most tables. ... View more

Hai.kuo definitely lived up to his name here, but... this question is not marked as "answered" yet, so I thought I'd provide one more option (similar to DBailey) Assuming that Product_type is a character variable data have;    infile datalines ;    informat Purchase_Date date9.;    format   Purchase_Date date9.;    length   Transaction_Ref $5 Product_Type $10;    input    customer_number  Purchase_Date Transaction_Ref Product_Type ;    datalines;             1                16JUL2014     12345           TYPE1             1                17JUL2014     12346           TYPE1             1                18JUL2014     12347           TYPE1             1                18JUL2014     12347           TYPE2             1                18JUL2014     12347           TYPE3             1                19JUL2014     12347           TYPE1             2                16JUL2014     22345           TYPE1             2                17JUL2014     22346           TYPE1             2                18JUL2014     22347           TYPE1             2                18JUL2014     22347           TYPE2             2                18JUL2014     22347           TYPE3             2                19JUL2014     22347           TYPE1 ; run; proc sql;  create table want as    select    a.customer_number             ,a.Purchase_Date             ,a.Transaction_Ref             ,a.Product_Type             ,count( DISTINCT                    (case when b.purchase_date between a.purchase_date-365                                                   and a.purchase_date-1                          then b.product_type                          else ''                     end)                    )    as type_ct    from      have   A    left join have   B   on a.customer_number = b.customer_number    group by  1,2,3,4 ;quit; ... View more

I submitted an idea with a similar name: SAS-EG Clear Results Programmatically Please vote ... View more

The attached project has only 4 output datasets, so I'm not reaching that limit. I purposely used a very simple example from a small dataset in the SAS Help library. ... View more

Additional details: I see the same behavior in version 5.1 of EG. The interesting thing is that the List_O program does open the way I want it to open if I select only the first 3 programs (skipping the "continue" program) If I select all programs (including the "continue" program) or choose the "Run Branch" or "Run Process flow", then List_O doesn't open automatically. ... View more

Jaap, I'm all in favor of improving SAS, but I'm not sure that SAS is doing anything wrong in this case. My best guess is that 8 bytes is not enough precision for 6 decimals in the date-time stamp, and 8 bytes is the largest size of a numeric variable. I would have to do a lot more testing before submitting this to SAS Tech Support. ... View more

Jaap, please run through Owen's original code and watch the decimals. From what I understand, Owen would like to skip the decimals (show all zeroes for the decimals) instead of showing incorrect decimals. That's why he would like to read pre-1926 and post-2095 dates with low precision(ymddttm19.) and contemporary dates with high precision(ymddttm26.) Even though SAS recognizes date-time stamps between 1582 and 1926, it recognizes them only to whole seconds, not to one-millionth of a second. On the one hand it would be great if SAS would allow 6-decimal precision for all date-time values, but on the other hand, do we really need to know that the "October Revolution" started exactly at '07nov1917 21:45:00.123456'dt? I think that '07nov1917 21:45:00'dt is close enough. Dates prior to 1582 could be off by 10 days, so I don't want to go there. ... View more