I have an application that is writing to a dataset using the table API (H5TB) using a compound type that contains another compound type, some scalars, some arrays, and two enums. When I try and look at the dataset using h5dump, it fails to resolve the enums and I'm not sure why.
The datatype the dataset is using appears to be not committed, that is, the structure definition is printed by h5dump right after the dataset section start, rather than simply referring to the path of a committed type. The enums in question are listed with the uncommitted type definition as expected.
For example, I see this in the datatype definition:
"STATION_POWER_STATUS" 5636;
but in the actual data, I see only:
00:00:16:04,
(5636 = 0x1604)
What might I be doing wrong to cause this disconnect for the enum types? I'm doing similar stuff with the low-level H5D API without issue.
Hi, Quincey,
Sorry, I should have posted a follow-up... in short, it turned out that I had managed to swap two enum type IDs somewhere and there was no overlap of values between the two. It's since been fixed.
···
On 04/04/2012 04:54 PM, Quincey Koziol wrote:
Hi John,
On Mar 30, 2012, at 4:10 PM, John K wrote:
I have an application that is writing to a dataset using the table API (H5TB) using a compound type that contains another compound type, some scalars, some arrays, and two enums. When I try and look at the dataset using h5dump, it fails to resolve the enums and I'm not sure why.
The datatype the dataset is using appears to be not committed, that is, the structure definition is printed by h5dump right after the dataset section start, rather than simply referring to the path of a committed type. The enums in question are listed with the uncommitted type definition as expected.
For example, I see this in the datatype definition:
"STATION_POWER_STATUS" 5636;
but in the actual data, I see only:
00:00:16:04,
(5636 = 0x1604)
What might I be doing wrong to cause this disconnect for the enum types? I'm doing similar stuff with the low-level H5D API without issue.
Hmm, that's odd... I'll CC one of our "tools guys" to see if he might know.
Allen - any ideas?
Quincey
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
Hi, Quincey,
Sorry, I should have posted a follow-up... in short, it turned out that I had managed to swap two enum type IDs somewhere and there was no overlap of values between the two. It's since been fixed.
Ah, very good, thanks!
Quincey
···
On Apr 4, 2012, at 5:05 PM, John K wrote:
On 04/04/2012 04:54 PM, Quincey Koziol wrote:
Hi John,
On Mar 30, 2012, at 4:10 PM, John K wrote:
I have an application that is writing to a dataset using the table API (H5TB) using a compound type that contains another compound type, some scalars, some arrays, and two enums. When I try and look at the dataset using h5dump, it fails to resolve the enums and I'm not sure why.
The datatype the dataset is using appears to be not committed, that is, the structure definition is printed by h5dump right after the dataset section start, rather than simply referring to the path of a committed type. The enums in question are listed with the uncommitted type definition as expected.
For example, I see this in the datatype definition:
"STATION_POWER_STATUS" 5636;
but in the actual data, I see only:
00:00:16:04,
(5636 = 0x1604)
What might I be doing wrong to cause this disconnect for the enum types? I'm doing similar stuff with the low-level H5D API without issue.
Hmm, that's odd... I'll CC one of our "tools guys" to see if he might know.
Allen - any ideas?
Quincey
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
