Good morning @rob
I wrinting this post because I was using QSDA for the last two years and I think it’s a very useful tool. It helps me a lot however, I have seen that it warns me about false positives regarding if a column is used or unused. Of course, that’s happening when the QS Data Model was created using the Data Manager GUI. You know, in these cases if we want to filter by any column then this column has to be included in the data model although it isn’t used in any chart. If we analyze our app with QSDA, it’s going to tell us that this column is not used however, it’s so, is planned to include any solution for this casuistry in any future release?
The same happens when we want to create a new column based on two, three, etc. columns, in this case if these columns aren’t used in any chart then the same story happens again.
That’s what I saw and I understand that QSDA can be running in this way by design so in this case, I’d like to know it. On the other hand, probably I’m not doing something well. I’d like to know it too.
Thanks in advance
When you say “if we want to filter by any column” do you mean using the Selections tool?
When you say “create a new column based on two, three, etc. columns,” do you mean a calculated column like:
Customer & ' - ' & City as "Customer Location"
Let me explain myselft:
Filters → when using Data Manager we filter the data set (inside WHERE sentences) → CustomerCountry = ‘France’. Although we won’t use CustomerCountry in any chart, Data Manager obliges us to include it in our data model then QSDA is telling us CustomerCountry is not used however it’s because it was used inside the WHERE sentence.
Create a new column based on two, three, … columns → You’re right, that’s what I meant.
- I also noticed that one column with a subfield create by AutoCalendar is used only in a “Background color expression” then QSDA says that the main column is not used. Let me show you:
If you need me to check anything else please let me know.
Ah! the new column formed by the combination of three columns was made like that: Column1&Column2&Column3
QSDA Field analysis only considers if a field is used in the UI. QSDA does not analyze whether a field is used in the script. If you use the DROP FIELDS statement, it should be placed at the end of the script, after all other script has completed.
Column1&Column2 as NewField
if only “NewField” is used in UI, “Column1” and “Column2” will be considered unused. Again, unused field analysis considers on UI (Front end) usage.
The issue with Derived fields like Date.AutoCalendar not being linked to the base field “Date” is a bug. I will address that.
It’s clear, althought as much as we work with Qlik finally we move to the Load Script Editor, the common path is to start with Data Manager and then move to the Script so, it’s something to take in mind when we use QSDA, I mean, it’s the ideal tool if you use the Script instead the Data Manager UI, else you can obtain false Un used results.
Anyway, I also found a new bug or I think so, it happens with release 2.5.0. If you don’t have any app analyzed and click over the buttón “Collected only” then the list of apps is emptied and the button dissapears so you have to refresh the browser to be able to see again the list of apps. In my opinion the “Collected only” button should be always visible.
Afther enabling the checkbox button:
The derived fields issue and the “Collected only” visibility issue are fixed in release 2.5.2.
Super! I’ll install it this evening.
Thanks a lot for fixing it so fast.
I installed the last release and every issue I mentioned here were fixed so thanks a lot.