[stgt] [PATCH 0/5 RESEND] mgmt and concat_buf

[FUJITA Tomonori]

On Tue, 14 Feb 2012 17:25:00 +0200
Alexander Nezhinsky <alexandern at mellanox.com> wrote:

> On 02/12/2012 01:01 AM, FUJITA Tomonori wrote:
> 
>> I got compile errors with this patchset. You can find my tree with
>> this patchset:
>> git://github.com/fujita/tgt.git concat
>> 
>> I got something like the following:
>> In file included from iscsi/iscsid.h:30,
>>                  from iscsi/transport.c:27:
>> ./tgtd.h:119: error: expected specifier-qualifier-list before
>>                  ‘tgtadm_err’
> 
> 
> Tomo,
> 
> I checked out your "concat" branch. The commit corresponding to the patch 2/5
> looks like this:
> 
> # git show 61bb9b27ddb9237bda3c08791ea97017139d18b5
> commit 61bb9b27ddb9237bda3c08791ea97017139d18b5
> Author: Alexander Nezhinsky <alexandern at mellanox.com>
> Date:   Wed Feb 1 15:42:46 2012 +0200
> 
>     mgmt and concat_buf: define tgtadm_err, add error values to tgtadm_error.h
> ...
> +       TGTADM_NO_BINDING,
>         TGTADM_TARGET_EXIST,
>         TGTADM_LUN_EXIST,
> +       TGTADM_BINDING_EXIST,
>  
>         TGTADM_ACL_EXIST,
>         TGTADM_ACL_NOEXIST,
> 
> Here it ends. If you look the patch 2/5 that i sent, it is concluded with the
> following lines:
> +typedef enum tgtadm_errno tgtadm_err;
> +
>  #endif

Oops, I can't cleanly apply that patch so I did by hand.

> And this typedef of "tgtadm_err" is exactly what's lacking according 
> to the compilation error message you'd sent. Whatever the reason that
> the patch was applied only partially, this is the cause of the errors.
> 
>> In addition, the first patch doesn't look 'git bisect'able. Please
> 
>> make sure every patch compilable.
> 
> 
> My fault, i did not pay attention to that.
> Unfortunately, not only the 1st patch breaks bisectability, actually
> you need to apply all 5 patches to get everything compilable.
> This happened because i produced them on the file-by-file basis after
> rebase. 
> 
> I don't see any easy way now to produce a fine-grained patch set.
> Thus i am going to send you a unified patch with all the changes.
> Not perfect but i agree that keeping bisectability is important.