hi Vidyaranya,

Please check my comments inline.


Thanks,

William


________________________________
From: Vidyaranya Malladi <[email protected]>
Sent: Friday, September 29, 2017 10:06 PM
To: [email protected]; [email protected]
Cc: [email protected]; Mara Preotescu; Abishek Kunduru; 
Ananthanarayanan Ms
Subject: Griffin Accuracy Doubts

Dear William,
 Good day. Please let us know on the below questions that we have

1. Does Griffin plan to support multi table being src eg. src table frame from 
a qry joined thro tabl1,tabl2,tabln and selected few cols and then finally use 
that as src and similarly for destination as well

--Yes, Since we have change our backend to fully support sql.

2. Would Griffin support the result metric col wise instead of overall matched 
count.
eg if src table is having 4 cols and dest is having 4 or more cols and user 
wants to check the accuracy of col4 which is 'price' in src and destination by 
having a new col added to the src table as '%change' which would be 
abs(src_price - dest_price)/100

--Yes, Since we have change our backend to fully support sql.

Fact_tag        fact_sDesc      fat_Ldesc       SRC Fact_precision      SRC 
Data        TGT Data        % Chg
f000000000000000000100000000000000000000        Units   Sales Units     1       
25510.00        2551.00 0.00
f000000000000000000100000000000000000001        Units (any promo)       Sales 
Units (any promo) 1       5846.00 584.59  0.00
f000000000000000000100000000000000000002        Units (no promo)        Sales 
Units (no promo)  1       17172.00        1717.25 0.00


3. Support udfs on the cols before comparing, eg one of our src is a table from 
mainframe format where dots/decimals are ignore and we have to apply a logic to 
get the decimals

--For UDFs, if it registered in your computing cluster, it should be fine from 
command line. we are considering to refactor ui to support sql editor directly.
but how to provide better user experience from front end for UDF, we are still 
considering it.

4. Last time we discussed few features were said to be discussed to support 
could you please let us know if we have any road map for the same. like 
advanced sql features and adding groupby clause/udfs etc

--Support SQL is already done, will have a quick release after we fully tested 
and reviewed.

5. Griffin compares the cols selected by using them in joins, do we plan to 
enhance to check if the user could select few cols other than the mentioned in 
joins and we need to compare the src col's value vs dest col's value lets say 
without the join and only by order by clause.


select col from src order by prikey
select col from tgt order by prikey

src.col vs tgt.col instead of the join key . Please let us know.

-- For this case, it depends on how would you like to map src.col with tgt.col, 
we don't need join key if you can elaborate mapping in sql since we are fully 
supporting sql.

Thank you very much.



Thanks & Regards
M. Vidyaranya
NDX
TCS - Nielsen Relationship
Mobile  : +91 9790591311

Reply via email to