More refactoring to completely (?) mimic previous library behavior for scanning contents of the file #11
Conversation
…ecial()
to clean up resources properly on failure.
Refactored doubly-linked list code for objects in the libsrc4 directory,
cleaning up the add/del routines, breaking out the common next/prev
pointers into a struct and extracting the add/del operations on them,
changed the list of dims to add new dims in the same order as the other
types, made all add routines able to optionally return a pointer to the
newly created object.
Removed some dead code (pg_var(), nc4_pg_var1(), nc4_pg_varm(), misc. small
routines, etc)
Fixed fill value handling for string types in nc4_get_vara().
Changed many malloc()+strcpy() pairs into calls to strdup().
Cleaned up misc. other minor Coverity issues.
…behavior of backwards ordering)
…ordering as the rest of the dimension queries. Correct error in library where types used in sub-group variables but that were added to the file after the sub-group was created weren't available for sub-group variables to use. Start cleaning up test suite and un-commenting tests that were commented out (got up to nc_test4/tst_fills2.c, alphabetically) before running into an error in HDF5.
|
Seeing an issue on 32-bit Windows builds in nc_test4/tst_converts2.c. The issue is recorded at the CDash site here: http://my.cdash.org/testDetails.php?test=13241998&build=551669 I can fix this easily enough for 32-bit Windows, but just wanted to make a note of it. |
|
On Dec 16, 2013, at 12:13 PM, Ward Fisher notifications@github.com wrote:
|
|
Sounds good; everything else looks fine re: our test platforms. Once Russ has made his review I'll merge it into the trunk. |
|
I've merged into the trunk; in addition to the windows issue mentioned above, we're seeing the same test failure on sol and my 32-bit Linux system, but this should be easily addressable. |
|
Looks like you've checked in a fix that's working for this, so I'm going to go on to the new Coverity issues. (And then back to working my way through the tests) On Dec 16, 2013, at 4:50 PM, Ward Fisher notifications@github.com wrote:
|
|
On 12/28/13, 12:23 PM, Quincey Koziol wrote:
-Ward |
|
On Dec 28, 2013, at 3:51 PM, Ward Fisher notifications@github.com wrote:
|
…h_warnings Ejh warnings + ward commit
Switch test for nc_inq_unlimdim and nc_inq_unlimdims to use the same ordering as the rest of the dimension queries. Correct error in library where types used in sub-group variables but that were added to the file after the sub-group was created weren't available for sub-group variables to use. Start cleaning up test suite and un-commenting tests that were commented out (got up to nc_test4/tst_fills2.c, alphabetically) before running into an error in HDF5.