[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: netcdf, OSF Alpha port

> Organization: RSI
> Keywords: 199403142107.AA14790

Hi Joshua,

> I am changing IDL to use NetCDF 2.3.2 instead of 2.1 or 2.2 -- I can't
> remember which.  In any case, it seems to work fine on my RS 6K but
> it seems to fail in ncvarput() on the Alpha complaining of "Invalid
> edge legnth"  line 622, NCvcmaxcontig in putget.c
> 
> The idl code which produces this effect:
> 
> id = ncdf_create('foo.cdf', /clobber)
> d0= ncdf_dimdef(id,'d0',10)
> d1 = ncdf_dimdef(id,'d1',20)
> vid = ncdf_vardef(id,'v1',[d0,d1],/byte)
> ncdf_control,id,/endef
> ncdf_varput, id, vid, bindgen(10,20)  ;; write 10x20 array of bytes
> 
> edp is a long (an 8 byte quantity) and ends up being 0x100000001 which
> just screams 4 byte / 8 byte assumption /reality conflict.
> 
> Have you done much Alpha testing?

Yes, in the sense that we have gotten the nctest program, which tests all
the netcdf library interfaces, running successfully on the Alpha.  However
we know there may still be bugs, because nctest doesn't test everything
exhaustively.

> If this is a bug how soon can I get a new copy.

We'll have to duplicate the problem here first.  Thanks for reporting enough
detail so that we can reproduce it.  I'm involved in an all day meeting, so
in case I don't get to it this afternoon, I'm also forwarding this to Steve
Emmerson (address@hidden) in case he has time to work on it today.

> I'm here in Boulder so if you want to sit down an run through a debug
> session with me I'd be amenable.

Thanks for the offer, but I don't think that will be necessary.

--Russ