Fix for table panic

[andschneider]

Description

This is an initial fix for issue #470.

After walking through one of the cases that caused the panic in a debugger, I found that it was due to a miscalculation (it seems like a rounding issue) in the column widths from the cassowary solver. I don't fully understand the parameters for this solver and as such I didn't feel comfortable trying to implement a fix into the get_column_widths function. Changing the solver also seems to have potential knock-on effects in how all tables are rendered, which sounds like a disruptive change.

So, I tried to make a minimal fix by simply checking for overflow in the set_style function and clamping the value if there would have been an overflow.

I'm not exactly sure if this is a completely robust solution, but it seems to work for the two test cases we have so far.

@fdehau would love your thoughts on how robust of a fix this is!

Testing guidelines

I added the two example tables that were causing a panic from the issue in widgets_table_columns_dont_panic test function. All this test does is render the tables and ensures a panic doesn't occur. It doesn't verify the output of the table.

Checklist

• I have read the contributing guidelines.
• I have added relevant tests.
• I have documented all new additions.

[andschneider]

So after a little more testing:

for i in 10..1000 {
    test_case(&mut state, table2.clone(), i);
}

I found that with table widths below 100, table2 from the tests still panics sometimes. Not sure why, but this fix is definitely not completely robust. It does work above 100 though, and table1 still does not panic.

[deepu105]

I'll test this branch as soon as I find some time

[deepu105]

This seems to work, atleast for the scenarios I tested. but regrdless its better than the current behaviour

[deepu105]

@fdehau WDYT? it would be nice to have this in a patch release

[fdehau]

Thanks for looking into it and sorry for the late response. I'm not very fond of the proposed fix because it can potentially hide a lot of bugs which will be harder to track and fix. Thanks to your provided regression test, I was able to identify an issue in the column widths computation. I've opened this PR which reuses existing code and should fix the issue.

[andschneider]

Thanks @fdehau! Glad there was a better solution 💯