a clean commit for #383 #731
Conversation
|
Looks good I will see how quickly I can get this in, we are trying to get 4.6.0 (previously 4.5.1, but with compression plugin support we have bumped to 4.6.0) out ASAP. I see the checks failed, will investigate in AM. |
| @@ -9,7 +9,9 @@ | |||
| #ifndef _DISPATCH_H | |||
| #define _DISPATCH_H | |||
|
|
|||
| #if HAVE_CONFIG_H | |||
There was a problem hiding this comment.
We always have config.h, right?
There was a problem hiding this comment.
This is a common practice when using autotools.
|
Awesome, thanks. Much better than the other PR! ;-) @WardF if you are going to include this in 3.6.0 then I think we will need a little time after this merge is complete to look into all the other reported fill value issues and make sure all is working as it should. For example I have some new tests that will need to be modified to check per-variable fill mode instead of per-file fill mode for classic formats. |
|
@wkliao after this PR is merged, what happens when the user calls nc_set_fill() to try and set fill value for the whole file. Does that still work? |
| @@ -158,6 +158,9 @@ NCD2_rename_var(int ncid, int varid, const char *name); | |||
| extern int | |||
| NCD2_var_par_access(int, int, int); | |||
|
|
|||
| extern int | |||
| NCD2_def_var_fill(int, int, int, const void *); | |||
|
|
|||
There was a problem hiding this comment.
Again, why was this moved? Seems like at least three files are changing for no real reason. Or am I missing something?
There was a problem hiding this comment.
OK, never mind. I see that you are moving things outside the #ifdef USE_NETCDF4. Now I get it. ;-)
|
OK, to answer my own question, after this PR, when nc_set_fill() is called it will correctly set the fill value of all variables. |
|
I am reviewing pull requests now and trying to prioritize them, I'd like to include this but we shall see, I also need to get 4.6.0 out to revert some behavior that was introduced in 4.5.0. So.. maybe. :) |
|
Thanks! |
This PR contains only my changes for #383.