My initial response went to Alison only, by accident; my apologies. I’ve
been able to observe the issue in a C program provided by Alison, so that
rules out a Fortran-specific cause.
I’ve opened a ticket on GitHub for this issue, viewable at
https://github.com/Unidata/netcdf-c/issues/233 . Searching the web for HDF5:
infinite loop closing library turns up a number of other instances of this
happening. Interestingly enough, when building a version of hdf5 1.8.16 for
debugging, the issue itself disappeared. I’m going to try to narrow down
which of the configure options (--enable-codestack, --enable-debug=all,
--enable-using-memchecker) resulted in the issue resolving itself.
I’ll be posting my findings at the GitHub issue linked above as well as
following up here once there’s a resolution, but the short answer (as has
already been pointed out) is that Alison isn’t doing anything wrong in the
test (although using redef isn’t a bad idea).
-Ward
On Tue, Mar 8, 2016 at 9:57 PM, rweed <rweed@xxxxxxxxxxxxxxxx> wrote:
Alison
>
> If you change
>
> write(unit=myattr,fmt='(a6,i4.4)') 'myattr',i
>
> to
>
> write(unit=myattr,fmt='(a6,i4)') 'myattr',i
>
> the error disappears
>
> after the write with your format, myattr
> looks something like myattr0001 etc
> If you replace i4.4 with i4 myattr looks
> like myattr 1 (spaces instead of 0's
>
> I'm not sure why this would make a difference but it does.
>
> RW
>
>
> On 03/08/2016 02:19 PM, Alison Walker wrote:
>
>>
>> Roy, thanks for your response. I tried adding the call to nf90_redef
>> after the open, as you suggest, and then re-write the attributes, but I
>> still get the same error.
>> Alison
>>
>> On 3/7/2016 9:36 PM, Roy Mendelssohn - NOAA Federal wrote:
>>
>>> Hi Alison:
>>>
>>> I am not certain of this, but nc_create and nc_open behave
>>> differently. nc_create creates the file and puts you in define mode,
>>> nc_open does not put you in define mode. I don’t know what is magic
>>> about 30 attributes, perhaps that is what it takes to exceed the
>>> global attributes space because the history changes, but I wonder what
>>> happens if after the nc_open you call NF90_REDEF and then rewrite the
>>> global attributes.
>>>
>>> As I said, I could be wrong, but it would be easy enough to test.
>>>
>>> -Roy
>>>
>>>
>>>
>>>
>>> On Mar 7, 2016, at 7:01 PM, Alison Walker
>>>> <postings@xxxxxxxxxxxxxxxxxxxxxx> wrote:
>>>>
>>>> Hi netcdf folks,
>>>> I'm getting this error when changing the values of global attributes
>>>> in a netcdf4 file using the fortran interface:
>>>>
>>>> HDF5: infinite loop closing library
>>>>
>>>> D,G,A,S,T,F,FD,P,FD,P,FD,P,E,E,SL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL
>>>>
>>>>
>>>> Here is a little test program that illustrates the problem. It first
>>>> creates a netcdf file and writes some global attributes, then tries
>>>> to write them again after closing and re-opening the file.
>>>> Interestingly, if the number of attributes is less than 30, there is
>>>> no error. If the number is 30, the error occurs. I'm using netcdf
>>>> 4.3.0, netcdf fortran 4.2 and hdf5 1.8.16 on Linux x86_64.
>>>>
>>>> program tester
>>>>
>>>> use netcdf
>>>> implicit none
>>>>
>>>> integer, parameter :: n_attrs = 30
>>>>
>>>> integer :: iok, ncid, varid, i
>>>> character*10 :: myattr
>>>>
>>>> ! Create the file and write global attrs
>>>>
>>>> iok = nf90_create('tester.nc',NF90_HDF5,ncid)
>>>> print *,'create iok=',iok
>>>> do i=1,n_attrs
>>>> write(unit=myattr,fmt='(a6,i4.4)') 'myattr',i
>>>> iok = nf90_put_att(ncid,NF90_GLOBAL,myattr,i)
>>>> print *,i,' put iok=',iok
>>>> enddo
>>>> iok = nf90_close(ncid)
>>>> print *,'close iok=',iok
>>>>
>>>> ! Open the file again and re-write global attrs
>>>>
>>>> iok = nf90_open('tester.nc',NF90_WRITE,ncid)
>>>> print *,'open iok=',iok
>>>> do i=1,n_attrs
>>>> write(unit=myattr,fmt='(a6,i4.4)') 'myattr',i
>>>> iok = nf90_put_att(ncid,NF90_GLOBAL,myattr,i)
>>>> print *,i,' put iok=',iok
>>>> enddo
>>>> iok = nf90_close(ncid)
>>>> print *,'close iok=',iok
>>>>
>>>> end program tester
>>>>
>>>> When I run this test program with 30 attributes I get this error result:
>>>>
>>>> create iok= 0
>>>> 1 put iok= 0
>>>> 2 put iok= 0
>>>> 3 put iok= 0
>>>> 4 put iok= 0
>>>> 5 put iok= 0
>>>> 6 put iok= 0
>>>> 7 put iok= 0
>>>> 8 put iok= 0
>>>> 9 put iok= 0
>>>> 10 put iok= 0
>>>> 11 put iok= 0
>>>> 12 put iok= 0
>>>> 13 put iok= 0
>>>> 14 put iok= 0
>>>> 15 put iok= 0
>>>> 16 put iok= 0
>>>> 17 put iok= 0
>>>> 18 put iok= 0
>>>> 19 put iok= 0
>>>> 20 put iok= 0
>>>> 21 put iok= 0
>>>> 22 put iok= 0
>>>> 23 put iok= 0
>>>> 24 put iok= 0
>>>> 25 put iok= 0
>>>> 26 put iok= 0
>>>> 27 put iok= 0
>>>> 28 put iok= 0
>>>> 29 put iok= 0
>>>> 30 put iok= 0
>>>> close iok= 0
>>>> open iok= 0
>>>> 1 put iok= 0
>>>> 2 put iok= 0
>>>> 3 put iok= 0
>>>> 4 put iok= 0
>>>> 5 put iok= 0
>>>> 6 put iok= 0
>>>> 7 put iok= 0
>>>> 8 put iok= 0
>>>> 9 put iok= 0
>>>> 10 put iok= 0
>>>> 11 put iok= 0
>>>> 12 put iok= 0
>>>> 13 put iok= 0
>>>> 14 put iok= 0
>>>> 15 put iok= 0
>>>> 16 put iok= 0
>>>> 17 put iok= 0
>>>> 18 put iok= 0
>>>> 19 put iok= 0
>>>> 20 put iok= 0
>>>> 21 put iok= 0
>>>> 22 put iok= 0
>>>> 23 put iok= 0
>>>> 24 put iok= 0
>>>> 25 put iok= 0
>>>> 26 put iok= 0
>>>> 27 put iok= 0
>>>> 28 put iok= 0
>>>> 29 put iok= 0
>>>> 30 put iok= 0
>>>> close iok= 0
>>>> HDF5: infinite loop closing library
>>>>
>>>> D,G,A,S,T,F,FD,P,FD,P,FD,P,E,E,SL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL,FL
>>>>
>>>>
>>>> Can anyone help me to understand why this error is occurring when
>>>> re-writing the attributes?is it a bug with netcdf/hdf or am I doing
>>>> something wrong?
>>>> Thanks,
>>>> Alison
>>>>
>>>> _______________________________________________
>>>> netcdfgroup mailing list
>>>> netcdfgroup@xxxxxxxxxxxxxxxx
>>>> For list information or to unsubscribe, visit:
>>>> http://www.unidata.ucar.edu/mailing_lists/
>>>>
>>> **********************
>>> "The contents of this message do not reflect any position of the U.S.
>>> Government or NOAA."
>>> **********************
>>> Roy Mendelssohn
>>> Supervisory Operations Research Analyst
>>> NOAA/NMFS
>>> Environmental Research Division
>>> Southwest Fisheries Science Center
>>> ***Note new address and phone***
>>> 110 Shaffer Road
>>> Santa Cruz, CA 95060
>>> Phone: (831)-420-3666
>>> Fax: (831) 420-3980
>>> e-mail: Roy.Mendelssohn@xxxxxxxx www: http://www.pfeg.noaa.gov/
>>>
>>> "Old age and treachery will overcome youth and skill."
>>> "From those who have been given much, much will be expected"
>>> "the arc of the moral universe is long, but it bends toward justice"
>>> -MLK Jr.
>>>
>>>
>>>
>> _______________________________________________
>> netcdfgroup mailing list
>> netcdfgroup@xxxxxxxxxxxxxxxx
>> For list information or to unsubscribe, visit:
>> http://www.unidata.ucar.edu/mailing_lists/
>>
>
> --
> --
> ---
>
> Richard Weed, Ph.D.
> Associate Research Professor
> Center for Advanced Vehicular Systems (CAVS)
> Mississippi State University
>
> Email: rweed@xxxxxxxxxxxxxxxx
> Ph: (662) 325-5450
>
>
> _______________________________________________
> netcdfgroup mailing list
> netcdfgroup@xxxxxxxxxxxxxxxx
> For list information or to unsubscribe, visit:
> http://www.unidata.ucar.edu/mailing_lists/
>
