Created attachment 29815 [details] problem what I faced
Can you provide a unit-test which shows the problem? It might be caused by processing before the few lines in the screenshot and thus might be hard to reproduce without this.
Created attachment 30359 [details] sample code
Created attachment 30360 [details] test case file
Created attachment 31823 [details] Afflicted fill entries from xl/styles.xml The first fill was preexisting in the file which was originally generated by Excel. The other two were created with POI. As per the Open XML SDK 2.0 Productivity Tool for Microsoft Office "The attribute 'rgb' has invalid value 'FF0000'. The actual length according to datatype 'hexBinary' is not equal to the specified length. The expected length is 4."
(In reply to Eric Riese from comment #4) > As per the Open XML SDK 2.0 Productivity Tool for Microsoft Office > "The attribute 'rgb' has invalid value 'FF0000'. The actual length according > to datatype 'hexBinary' is not equal to the specified length. The expected > length is 4." Some .xlsx colours require a 4 hex form (ARGB), others require a 3 hex form (RGB). At first glance, it would appear that the wrong kind is getting written Any chance someone could step into the colour and style code with a debugger, and see what is getting written where and why, to try to narrow down where the bug is coming in?
Created attachment 31851 [details] less a patch and more of a hint where the problem starts I was having trouble building and debugging this, but this is the least of what's wrong. This is the first place where the alpha byte is lost.
We improved support for transparency/alpha-channel via r1907106 and r1907105 which hopefully also solves this issue.