[PATCH 3/4] winegcc: Pass tool by id to build_tool_name.

[Francois Gouget]

On Wed, 4 Mar 2020, Marvin wrote:
[...]
> === debian10 (build log) ===
> 
> error: patch failed: tools/winegcc/winegcc.c:314
> Task: Patch failed to apply
> 
> === debian10 (build log) ===
> 
> error: patch failed: tools/winegcc/winegcc.c:314
> Task: Patch failed to apply

So what appears to have happened here is that, for whatever reason, part 
2/4 of the patch series got lost. This has caused a cascading failure 
for all of the 4-parts series you sent since:

Date: Mon, 2 Mar 2020 15:52:27 [PATCH 1/4] winegcc: Lookup PATH in 
Date: Mon, 2 Mar 2020 15:52:54 [PATCH 3/4] winegcc: Pass tool by id
Date: Mon, 2 Mar 2020 15:53:31 [PATCH 4/4] winegcc: Add support for 

-> Part 2 is missing so the patch series remained 'open' waiting for it.
   To not have such zombie patch series the TestBot deletes patch series 
   when the last patch is more than 24 hours old. The check is performed 
   once a day so the 24 hours were not up yet when the script ran on 
   Monday.


Date: Tue, 3 Mar 2020 14:01:59 [PATCH 1/4] winegcc: Lookup PATH in 

-> Since the patch series was still open and already had a part 1 this 
   patch got ignored and reported as 'Existing part reply / collision'.

   What should have been done here is mark this series as 'v2' but in 
   practice it would not have made a difference because the TestBot does 
   not yet take versions into account [1] :-(  (working on it)

Date: Tue, 3 Mar 2020 14:02:08 [PATCH 2/4] winegcc: Return build_tool

-> This patch completed the first patch series, causing it to 
   disappear. Miraculously the patch applied fine too. But of course it 
   did not test the right code.

Date: Tue, 3 Mar 2020 14:02:18 [PATCH 3/4] winegcc: Pass tool by id to
Date: Tue, 3 Mar 2020 14:02:28 [PATCH 4/4] winegcc: Add support for 

-> And those to created a new pending patch series waiting for parts 1 
   and 2.
   And on Tuesday night the latest patch in the series was still less 
   than 24 hours old.


Date: Wed, 4 Mar 2020 17:15:22 [PATCH 1/4] winebuild: Try also llvm 
Date: Wed, 4 Mar 2020 17:15:34 [PATCH 2/4] winebuild: Add an option to 

-> Those two completed the Tuesday patch series. So parts 1 and 2 tested 
   out fine but of course Tuesday's parts 3 and 4 did not apply cleanly 
   on top of Wednesday's parts 1 and 2.

Date: Wed, 4 Mar 2020 17:15:50 [PATCH 3/4] makedep: Use winebuild to 
Date: Wed, 4 Mar 2020 17:15:58 [PATCH 4/4] configure: Try using explicit

-> And these created yet another pending patch series which is going to 
   cause trouble and will not disappear before late on Thursday or 
   Friday morning.


So my recommended workaround would be for the next series to not have 4 
parts, or make a 48 hours pause (changing the email address would work 
too but that impacts the git record).


[1] Support for versioned patch series is the next database schema 
    update after the MissionCaps patch. I just modified handling of 
    patch series replies and I think now the patch series is ready.

-- 
Francois Gouget <fgouget at codeweavers.com>