Re: heads up on table change

On 18/03/2016 17:15, Glenn Linderman wrote:
> On 3/18/2016 9:25 AM, Robin Becker wrote:
>> In version 3.3.2,
>>
>> https://bitbucket.org/rptlab/reportlab/commits
>> /967759f7f778f1c03c6a8e1e248a525a9072a003
>>
>> a bug fix found by Vytis Banaitis (vytis @ bitbucket) removes double padding
>> in Table._calcPreliminaryWidths which can have effects on table column sizes.
>>
>> I looked at differences in all the reportlab test cases and there were only a
>> few. Personally I preferred the earlier outcomes, but using the padding twice
>> is clearly wrong.
>>
>> If you want to restore the earlier outcome then shout loudly here and I will
>> add some settings mechanism to accomplish that.
>



> Would it not be simple, for folks that want the earlier outcomes, to double
> their padding width?  Or would that have an effect on the outer padding of the
> outer columns that is different from before? In other words, you didn't really
> describe the effect of the bug completely, but the code you linked to makes it
> look like simply doubling the padding would recreate the original effect.
>
........
Unfortunately, I am almost certain it would not work. The left right+padding got 
added twice into the width estimation for columns without a specified width.

Later during layout the padding was used correctly. It ought to have more effect 
on smaller columns and my observations of the changes in the outcomes of 
test_platypus_tables seem to bear that out. The column width estimation in 
platypus tables seems a bit ad hoc to me so there is room for improvement.
-- 
Robin Becker