IDL messes up Julian day plots ?

[LC's No-Spam Newsreading account <nospam(at)ifctr.mi.cnr.it>]

I have a 2-year-long list of samples, one sample every 15 min.
The times are stored in Unix format (integer seconds since 1970).

I want to plot this using label_date to annotate the x axis. I do 

; convert to julian days
times=double(a.data.time)
times=times/86400. 
times=times + julday(1,1,1970)

dummy=label_date(date_format='%D %M')
plot,times,a.data.nitem   ,xtickformat='label_date'

If I restrict the plot to shorter periods e.g.
plot,times(u),a.data(u).nitem   ,xtickformat='label_date'

I get that several points are plotted at the same abscissa (x coordinate)
even if the values of a.data.time and times are distinct.

It looks like the plotting routines are unable to scale double precision
values correctly. On the other hand I'd say I need such precision with "large"
numbers like dates in julian days.  And I need julian days to use label_date.

Any clue ?

----------------------------------------------------------------------
nospam@ifctr.mi.cnr.it is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.