Lynton, > this is some progress. I will be interested to hear the outcome. In much > of my code development > debugging work I use Totalview. Do you have access to this or to > similiar software? I don't have access to TotalView, but use gdb from the emacs. I can trace the problem down to the recursive call nested three deep of H5T_convert() when the "datatype conversion failed" error occurs in trying to convert a vlen, but that doesn't mean it's necessarily an HDF5 error. I don't know of a practical way to determine that without writing the equivalent data using just HDF5 C calls, and that will take some concentrated time that I hope to find in the next week ... --Russ > Lynton > > On 09/07/2012 09:14 PM, Unidata netCDF Support wrote: > > Lynton, > > > > The problem has now been simplified to a small C program (attached) that > > creates a netCDF-4 file with one variable and gets an HDF5 error. It's > > not clear yet whether this is caused by a netCDF-4 bug or an HDF5 bug, but > > we should be able to determine that by testing an equivalent HDF5 program > > that doesn't use any netCDF-4 calls. That has yet to be done ... > > > > --Russ > > > >>> New Client Reply: Problems with CDL/vlen example > >>> > >>> thanks Russ, > >>> this is great news. I will need to schedule some time to get back to > >>> finish t > >>> he work.I cannot do it right away but will get back to it as soon as I > >>> can. > >>> Lynton > >>> > >>> -----Original Message----- > >>> From: Unidata netCDF Support [mailto:address@hidden > >>> Sent: 29 August 2012 20:55 > >>> To: Appel, Lynton > >>> Cc: address@hidden; address@hidden; Appel, Lynton; sa > >>> address@hidden > >>> Subject: [netCDF #JNH-860762]: Problems with CDL/vlen example > >>> > >>> Hi Lynton, > >>> > >>>> Dear Sylwester and Russ, > >>>> thanks for the enquiry > >>>> I spent a lot of time on developing an upgraded C++ API that could > >>>> handle user-defined in a much better way. The validation process has got > >>>> somewhat stalled by a bug in the netCDF-C layer or HDF5-layer. > >>>> Russ is there any news on this? > >>> Yes, there is as of this afternoon. The bug that was holding things up > >>> appears to have been fixed with changes to netCDF-4 or HDF5. I say "or", > >>> because at this point, it's difficult to determine whether the fix was in > >>> the upgrade from HDF5 1.8.8 to 1.8.9 or the netCDF-4 upgrade from 4.1.3 > >>> to 4.2.0, but it works now with the combination netCDF-4 version 4.2.1.1 > >>> and HDF5 version 1.8.9. The details are in this Jira issue description > >>> and comments: > >>> > >>> https://www.unidata.ucar.edu/jira/browse/NCF-155 > >>> > >>> Please try compiling and linking against the new versions and verify > >>> whether > >>> you agree that the bug that was blocking release of your new version of > >>> the > >>> netCDF-4 C++ release is now fixed. I'm hoping no new bugs affecting that > >>> code have been introduced ... > >>> > >>> --Russ > >>> > >>> > >>> > >>> Russ Rew UCAR Unidata Program > >>> address@hidden http://www.unidata.ucar.edu > >>> > >>> > >>> > >>> Ticket Details > >>> =================== > >>> Ticket ID: JNH-860762 > >>> Department: Support netCDF > >>> Priority: Critical > >>> Status: Closed > >>> > >>> > >>> > >>> Ticket Details > >>> =================== > >>> Ticket ID: JNH-860762 > >>> Department: Support netCDF > >>> Priority: Critical > >>> Status: Open > >>> Link: > >>> https://www.unidata.ucar.edu/esupport/staff/index.php?_m=tickets&_a=vi > >>> ewticket&ticketid=19304 > >> > > Russ Rew UCAR Unidata Program > > address@hidden http://www.unidata.ucar.edu > > > > > > > > Ticket Details > > =================== > > Ticket ID: JNH-860762 > > Department: Support netCDF > > Priority: Critical > > Status: Closed > > Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: JNH-860762 Department: Support netCDF Priority: Critical Status: Closed
NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.