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

Re: Deployed: Bug fixes to NetCDF Java API 4.1.20100227.0121

Thanks, John.  You and I can take the discussion "off-line" (i.e. we can
just email each other).  I'll send you another email with specific
comments etc. and we can go from there.
> On 3/5/2010 5:07 PM, Tom Margolis wrote:
>> Hi,
>>
>> I found and fixed bugs - having to do with the handling of longitude
>> ranges - in the following two NetCDF API classes:
>>
>>      ucar.nc2.dataset.CoordinateAxis1D
>>      ucar.nc2.dt.grid.GridCoordSys
>>
>> I did so by adding a new ucar.ral.util.BugFixes class, and having the
>> above two classes call methods therein.  This isolation will make it
>> easy to find and identify the bug fixes: search for the string
>> "BugFixes" in the above two classes.
>>
>> I jarred up the original 4.1.20100227.0121 code base with the two
>> altered classes and the new BugFixes class, and deployed to the MIT
>> Lincoln Labs repository as:
>>
>>
>> edu/ucar/unidata/netcdf-api/4.1.20100227.0121-ralfix.1/netcdf-api-4.1.20100227.0121-ralfix.1.jar
>>
>>
>>
>> The "ralfix" jar is also attached to this email, as are the three
>> classes in question.  The "ralfix" jar includes a CHANGES_ralfix file
>> that explains the changes made.
>>
>> John:
>>      Can you take a look at the fixes and possibly include in your next
>> release?
>>
>> Others:
>>      Please update your wcsri/ and wcsri_datatests/ code from svn.  The
>> updates will include new pom.xml files, new test code, and new core
>> code.
>>
>> Aaron:
>>      Can you pass this email to the folks at Savoir?
>>
>> Thanks,
>> Tom
>>    
>
> Hi Tom:
>
> I appreciate your time in tracking down these bugs. Sorry it takes me
> a while to review them, esp when they have implications throughout
> larger pieces of code.
>
> 1) CoordinateAxis1D
>
> The problem here is that the longitude axis is violating this
> assumption of CoordinateAxis1D:
>
>    * A 1-dimensional Coordinate Axis. Its values must be monotonic.
>
> It would be good to relax that assumption for longitude if possible,
> since obviously datasets normalize their longitudes.
>
> However, im not sure where else that assumption is used, so i would
> have to examine the grid code base. I wonder if youve thought about
> that and looked through the grid code for that assumption?
>
> Its possible that a better strategy is to have CoordinateAxis1D
> monoticize its values, ie check when reading in if it crosses the
> dateline and adjust.
>
> 2) GridCoordSys
>
> These appear to be the same as what you previously sent? If so, Ive
> already incorporated these in the release.
>
> It would be helpful to me in reviewing your fixes if you would
> summarize the problem, as well as the fix. A sample file (remote
> access ok) and/or a unit test would sometimes cut down on the work i
> have to do.
>
> thanks again.
>