test logs unnecessary filled with wrong data for 'low-precision-defined' parameters, affecting column width in *.xls round trip tests.
In the detailed test logs ( e.g. t6517-names.pl.log after 'make check TESTS' in the subdirectory test ) it teems with entries like:
<gnm:Cols DefaultSizePts="48">
- <gnm:ColInfo No="0" Unit="62.2**5**" HardSize="1"/>
- <gnm:ColInfo No="2" Unit="74.2**5**" HardSize="1"/>
+ <gnm:ColInfo No="0" Unit="62.2**6**" HardSize="1"/>
+ <gnm:ColInfo No="2" Unit="74.2**6**" HardSize="1"/>
as far as i understand after save to and then re-read from an xls file e.g. the entry 62.25 for width of col. A is changed to 62.26 for future saves ( at least to xml ).
As the correct value 62.25 is! available and shown in sheet / UI for Column - Width I'd propose to use that for file write, it would avoid a valuable amount of entries now spamming the logs.
evtl. the problem is limited to HiDPI systems which get fractional values for 'Column width in points'.
@aklapper i have! written a detailed description, IMHO it qualyfies as TL;DR and the active dev's here like it short, can send you the long version on request.