Calculations/filters go blank every time files are overwritten

Hi ,
I have a decently complex transformation(I think).

  • The three input files are updated from ERP manually .
  • I ensure that files have the same name/same sheet
  • The file structure , headers, columns are constant
  • After overwriting the files , whenever I open EDT the calculations & filters go blank

I would like to troubleshoot what I am doing wrong. I can email the data file to you if needed.

PO Wise Dashboard.transform (26.4 KB)

I have the feeling there is something corrupt in the stored file. After opening it, I can select some of the transformations and on some EDT crashes, which I cannot remember to have before. Maybe you need to create the transform file again, but I think Andy has more possibilities to analyse.

You are right , upon re-downloading the file from forum , it keeps crashing for me too. I am uploading from Windows machine now directly where it doesn’t crash.

PO Wise Dashboard.transform (26.4 KB)

Hi @prashant,

Without data it is hard to know where could be the issue, as if you click on any transform, you can’t see what it had and where the change could be happening.

I have a hunch, it could be due to auto run of transform


Change from auto run to manual and hope you will not have issues. The reason could be, is that when you are overwriting the files, the auto run kicks in and the files which are in the process of overwriting could not be read and hence you are loosing the setting in the transforms.

I will take a look when I get chance.

1 Like

The 2 .transform files linked above are identical.

I can’t really investigate without the corresponding input files. If they are sensitive it might be enough just to send the first few rows. Or you could use a Hash transform to encrypt sensitive columns.

I took out the one which kept crashing

Will Email you.

Just a thought, why not create a transform called Randomize and user can select which column values to randomize and it will randomize the content of the columns in place, without creating new column.

That is pretty much exactly what Hash does. Hashing is almost impossible to reverse, so can be used to anonymize data.


What I meant by randomize is that for example, if the column is date, it will put random dates, if it is number, it will put random numbers, if it is character field let say like names, it could take the names from the column and simply swap them around with each other, this way the data is still easy to read and understand and to use between transforms, and users could share their data which could be used to solve the issues they are facing.

Hmm or they can simply email you real sample and you solve their issues. So I guess no need. I was just thinking, if users could share their data easily, it really helps to understand what they are trying to achieve, as most of the time it’s just text trying to explain without test data and the outcome they are looking for, so it becomes hard to help or to come up with the solution.

We have considered generating randomized datasets as a possible future feature. Generating random data with similar characteristics as existing data would be significantly harder to do. And changing the letter order would not be sufficient randomization for confidential text data.

Also sometimes you need the exact data to reproduce the problem.

Any progress? like to know, what was causing filters go blank.

  • I had an excel file with images inside ,converting It to CSV resolved my issue.

  • As per @Admin testing even images shouldn’t be problem but I overwrote EDT files multiple times and couldn’t reproduce.