> On Wed, 9 Feb 2011, Dmitrii Dimandt wrote: >> So I guess that the problem is probably inherent in Excel itself and I pity >> Apple's develpers for getting this right. >> >> Because, on top of all things, 01/10/10 (which is October 10th, 2010 over >> here in Europe) is 40450 days from January 1st, 1900. > > There's some nasty stuff you can do to make excel default to dd/mm/yy instead > of mm/dd/yy, but generally it tries to guess on the locale and often gets it > wrong > >> *Not* 40452 days. According to Python at least. This is so weird, to say at >> least :( > > Microsoft got it wrong though - they think thought that there was a 29th of > Feb 1900, but there wasn't... Take a look at > http://svn.apache.org/repos/asf/poi/trunk/src/java/org/apache/poi/ss/usermodel/DateUtil.java > for the gory details of how to do it. I think the extra day is an > inclusive/exclusive thing. >
Oh. My. God :) > > In terms of Tika and excel formulas, an excel formula and a cell with the > same number in it ought to be rendered the same, assuming the same format was > applied to both cells. If you can find a test case where this isn't the case, > then please create a new bug in jira (ideally with a unit test!) and I'll > take a look > > Nick I don't promise anything very soon or very quick, but I'll try to, since I happen to have such an excel file on hand :)
