Get Data / PDF - 1st Table is missing last 2 columns

[mike_honey]

Data in this PBIX was read from the PDF: "FSSI and Non-FSSI Stock Status Table"

https://www.fisheries.noaa.gov/national/population-assessments/fishery-stock-status-updates

Data was loaded as separate queries, for Page001 & 2, and Table001 & 2.

As shown in the PBIX repro (link below), the Table001 query does not return the last 2 columns of data from the PDF. This is unexpected as the Page001, Page002 and Table002 queries all return a complete set of columns.

https://1drv.ms/f/s!AmLFDsG7h6JPgo0w5EaiTkQ2DBU2Fw

[v-qiuyu-msft]

Hi @mike_honey,

I have reported this issue internally: CRI 94945810. Will keep you update once I get any information. 

Best Regards,
Qiuyun Yu 

[Ehren]

Hi Mike. Thanks for reporting this. We're looking into it.

Ehren

[mike_honey]

Possibly related - I updated to Power BI Desktop 2.65.5313.621 64-bit (December 2018).  Now when I open that file, open the Query Editor and hit Refresh Preview, the 2 missing columns are included for Table001.

The bad news is that most of the table headers are totally messed up - where they were correct for Table002 and correct (if incomplete) for Table001, now they are spread across the first 8 rows of the result, seemingly at random.  Similar result from the Page results, so +1 for consistency I guess ...

It's now totally unusable, so I'm left really scrambling on this project.  I'll probably have to switch to Tabula or similar.

[Ehren]

Hi Mike. I apologize for the fact that the results are worse in the December release. As a preview connector, it's still a work in progress, and your feedback will help ensure it works better in this and other similar scenarios in the future. In the meantime, you may want to revert back to the November release, or use another PDF tool as you mentioned.

Ehren

[v-yuta-msft]

@mike_honey ,

This issue has been fixed currently.

Regards,

Jimmy Tao

[mike_honey]

What do you mean by "currently"?

I'm running Power BI Desktop 2.84.701.0 64-bit (August 2020), and when I open the repo I see a third results - the column headings in the Table001 query now have embedded carriage returns inside the column names, e.g:

Overfishing?
(Is Fishing
Mortality
above
Threshold?)

It makes it really hard to build reliable solutions on this component when it's results keep on changing, for the same input file.  If you are going to tinker with it, it should be clearly announced in the Power BI blog.

[Ehren]

Hi Mike. Are you re-importing the file, or opening an old pbix? We've tried to only make breaking changes behind a different "Implementation" option value. The old behavior (Implementation=null) should remain stable, unless it was due to an outright bug.

Ehren

[mike_honey]

@Ehren - thanks for exposing that, I had no idea it existed and there are no clues in the UI. Are there any plans to expose all those options in the UI?

I found the documentation but it seemed quite vague to me. What are the differences between implementations? When did "1.1" become the default?

It doesn't look like this change was announced. Will future changes be announced in the Power BI Desktop blog?  

It seems none of the usual PQ bloggers are aware of this either - any plans to let them know?

It's good to hear this component is being improved, but it's almost counterproductive if the changes aren't announced and described - that just leads to confused, frustrated users.

[Ehren]

Hi Mike. Thanks for the feedback. We don't plan to expose it in the UI, because the intention is for everyone to use the new Implementation going forward. The only reason for maintaining the old Implementations is for back-compat. The new Implementation includes various bug fixes, but they are fixes which could have broken existing mashups if they were applied across the board.

That said, I'll see if we can make more noise about it the next time we release a new Implementation.