Browser Out of Memory Error on first screen

[GoogleCodeExporter]

When loading a particular CSV file of around 100k rows, GR gets tricked into 
thinking that it should show the rows as records.
This causes the first state of the webapp to load with several hundred loaded 
rows in the grid, which subsequently kills the browser.

I could not escape this in any other way than doing some crazy browser tricks 
and then clicking "rows" instead of records.

GR: 2.0-r1836
Server: Custom Linux
Client: Mac OSX - Chrome

Original issue reported on code.google.com by aldo.buc...@gmail.com on 2 Dec 2010 at 6:07

[GoogleCodeExporter]

I presume your CSV is missing the values for the first column in a lot of cells?

Just as a work around, I would suggest: importing the file with "split into 
columns" turned off. Then transform the cells in that single column so that it 
contains something in the first column. For example, if the column separator is 
comma, then apply this transform: if(value.startsWith(","), "-" + value, 
value). Then split that column up.

Original comment by dfhu...@gmail.com on 2 Dec 2010 at 6:58

[GoogleCodeExporter]

Yes it does have batches of empty cells on the first col ;)

I found a way around it by modifying the source. Thanks for the tip.

Now, I would suggest that you implement some sort of limit so as to prevent the 
browser from committing suicide. This may happen in other scenarios. I guess.

Thanks!
A

Original comment by aldo.buc...@gmail.com on 2 Dec 2010 at 7:06

[GoogleCodeExporter]

Yes, the auto guessing of rows vs. records is too smart for its own good. I'm 
not entirely sure how to solve it right now, but it sounds like a valid issue.

Original comment by dfhu...@gmail.com on 2 Dec 2010 at 7:17

• Changed state: Accepted

[GoogleCodeExporter]

Perhaps we should have a limit on the number of rows per record which are 
displayed?  The XML importer bug which was causing records to be merged 
together triggered this behavior as well because, although Refine was only 
displaying two records, the second record was thousands of rows (effectively 
defeating the paging mechanism).

Original comment by tfmorris on 21 Oct 2011 at 4:15

[GoogleCodeExporter]

I've recently discovered that this condition can be triggered after the project 
has been imported if you blank out cells in the first column as part of a cell 
transformation operation.  The workaround was to move the column to the right 
so it didn't trigger the record aggregation, but we probably need to think 
about a way to warn the user when this happens.

Original comment by tfmorris on 16 Jan 2012 at 10:03