Hi,
1) negative julian dates SET(4, "yyyy-mm-dd") dDate := CTOD("") - 1 ? dDate // Clipper: "1449-02-15", Harbour: " - - "... So at beginning we should not expect that the date converted to string should will be non empty if date is not empty. It isn't even with Clipper. ? ctod("")+1 In this example we can only see that for some range of negative date values Clipper gives non empty date strings. Of course it's a bug. Probably dates are converted in some cyclic periods to non empty values. If you are interested then you can try to find this period. Anyhow it's a bug and should not be replicate. Please note that it's not revertable conversion, f.e. try this code:? (ctod("")-1) ? (ctod("")-1)-ctod("") ? ctod("1449-02-15")-ctod("") ? (ctod("")-1)==ctod("1449-02-15") If you begin to dig deeper in Clipper date conversions you will find also other bugs. For me current Harbour behavior is perfect and systematic without any unexpected anomalies like above.
Yes, I know these issues. I've looked to code before writing a letter. To be Clipper compatible we need to used unsigned long (we have signed now) for Julian date and use unsigned short (we have signed int now) for intermediate year representation. In conversions Julian <-> (year,month,day). Since this is not a few line change, I wanted to consult the mailing list about these issues. The main annoying thing is that: EMPTY(dDate) is true, but EMPTY(CTOD(DTOC(dDate))) is false, for dates like CTOD("")-1 in Harbour.
Since, this annoying thing is also valid for CTOD("")+1 for both Clipper and Harbour, let's leave it as it is.
Regards, Mindaugas _______________________________________________ Harbour mailing list Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
