Drill-Through with greater than 10,000 rows shows incorrect results
I have loaded CSV data composed of a little over 106,000 rows. An ID field is unique in each row. When I drill-through on a bar chart column that represents 90,000 rows the detail page is displayed. Each row includes the ID column and each row has a unique value for ID until I reach row 10,000. At this point, in each subsequent row the ID value is replaced by the value of the ID value in the first row of the list.
I am using Flexmonster v2.320
Thank you for the detailed description. It really helped us to reproduce the issue on our side. The fix will be delivered with minor release version 2.403, ETA Sep 11 (the fix will be available with version 2.324 as well). Please let me know if you have further questions.
Thanks for addressing this so quickly. I did notice an additional, but related issue. The drill-through with the issue I described was on the last bar in the bar chart. If I drill-through on other bars with smaller number of rows every row is repeating the same ID value (which again happens to be the same one that is repeated in the larger bar drill-through. Your fix may have handled both situations, but I wanted to make sure you were aware of this other possibly related issue.
We tested the component once again and did not manage to reproduce the second issue. However, we agree with you that it seems related to the first issue. By the way, it is reproducible not only in charts mode but on the grid as well. We will fix the first issue and ask you to check whether the second one is fixed as well. Does it work for you?
Yes. On a different matter. I am having an issue trying to use the Compressor. I posted my issue under the topic “Javadoc for the Compressor”
This is probably related, but does not seem like the exact same issue. For us, when loading large data sets (our current size is 160,000 rows), the drill down reports wildly varying results for some measure fields. The aggregated values seem correct, but the values for some but not all of the measures specifically seem to be duplicated or just wrong.
It can somewhat be seen in your 1 million row demo: http://www.flexmonster.com/demos/one-million-rows/
Every drill-down I try on that page results in an empty table. This is a pretty significant issue for those using our pivot table. IF it can’t be fixed sooner than September 11th, is there an older version of flexmonster that we can revert to where the issue is known not to exist?
Thank you for reporting this issue. If it is possible, please send us your sample data via email. This would greatly help us while testing all the fixes connected to drilling down.
As to your request about some older versions, it seems that these versions contain the issue as well. Please let us know in case you have additional questions.
Unfortunately, we can’t really share our data due to its sensitive nature. Does the million row sample not confirm the problem? If you are unable to reproduce the issue, we can try to generate some sample data that would demonstrate the problem, but it would take us some time.
Thank you for the answer.
Using our one million row demo we can only reproduce empty drill-through view.
As for the duplicated or wrong data, it is better to have a sample data because we did not manage to reproduce such behavior.
Also, it is highly possible that these issues will require separate fixes.
Looking forward to hearing from you.
Unfortunately, we are currently working under deadlines, so we don’t have time to try to create mock data to duplicate this issue for you, and then wait for a fix. If it helps, here is some information on our current report:
We are loading 160,000 rows, with 17 fields, and 7 of those being measures. The fields that aren’t measures are mostly strings. The measures are all numbers, with 6 out of the 7 being decimal numbers. In the drill-through, only the measures seem to be duplicated/wrong. There is one measure that doesn’t seem to suffer from any issues. For what it’s worth, it’s one of the two measures that are active, and it comes first in the json first row and in the measures list. It’s also the only integer measure.
If that information is not enough, or you aren’t able to create dummy data yourself and duplicate this, then I or a member of my team can probably circle back around within a couple weeks to answer any questions.
In the meantime, we are trying to work-around the issue for our customers. I’m trying to implement a custom drill-through in a bootstrap modal dialog, but I’m having an odd issue with it. In the long-run we probably prefers a custom drill-through anyway because our users want to be able to export data from it, but we weren’t expecting to have to implement it at this stage. With that being said, if you could take a look at the forum post I just created, it would probably be more valuable for us: http://www.flexmonster.com/question/cant-draw-additional-pivot-inside-bootstrap-modal/
Thank you for such a detailed description of your data source structure. We will spend some time investigating and update you with the results. Does it work for you?
Hello Jaime and Dennis,
I am glad to inform you that the minor release 2.403 is available for download now. The issue when drill-through with greater than 10,000 rows shows incorrect results was fixed.
You are welcome to update the component.