| Message ID | 20240521-fix-check-deps-v1-2-d149bf40e7ee@bootlin.com (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
| Series | get_info and check_deps fixes | expand |
On Tue, May 21, 2024 at 04:41:38PM GMT, Louis Chauvet wrote: > When using the commit strategy for the cover letter, the first commit is > empty. This commit can't be applied properly with git am, so just skip it. This shouldn't be required, because get_prep_branch_as_patches will already omit the cover letter commit: ez.py: 1408: strategy = get_cover_strategy() ignore_commits = None if strategy in {'commit', 'tip-commit'}: cover_commit = find_cover_commit() if cover_commit: ignore_commits = {cover_commit} Something else must be going on, maybe? -K
Le 21/05/24 - 11:20, Konstantin Ryabitsev a écrit : > On Tue, May 21, 2024 at 04:41:38PM GMT, Louis Chauvet wrote: > > When using the commit strategy for the cover letter, the first commit is > > empty. This commit can't be applied properly with git am, so just skip it. > > This shouldn't be required, because get_prep_branch_as_patches will > already omit the cover letter commit: > > ez.py: 1408: > > strategy = get_cover_strategy() > ignore_commits = None > if strategy in {'commit', 'tip-commit'}: > cover_commit = find_cover_commit() > if cover_commit: > ignore_commits = {cover_commit} > > Something else must be going on, maybe? > > -K Without my patch I have this log: Checking message-id: <20240516-yuv-v8-0-cf8d6f86430e@bootlin.com> Breaking thread to remove parents of 20240516-yuv-v8-0-cf8d6f86430e@bootlin.com Checking message-id: <20240516-b4-new-color-formats-v1-0-74cf9fe07317@bootlin.com> Breaking thread to remove parents of 20240516-b4-new-color-formats-v1-0-74cf9fe07317@bootlin.com Checking message-id: <20240516-writeback_line_by_line-v1-0-7b2e3bf9f1c9@bootlin.com> Breaking thread to remove parents of 20240516-writeback_line_by_line-v1-0-7b2e3bf9f1c9@bootlin.com Checking base-commit: drm-misc/drm-misc-next Testing if all patches can be applied to drm-misc/drm-misc-next [PATCH v8 1/17] drm/vkms: Code formatting [PATCH v8 2/17] drm/vkms: Use drm_frame directly [PATCH v8 3/17] drm/vkms: write/update the documentation for pixel conversion and pixel write functions [...] [PATCH 2/3] drm/vkms: Add a macro for write_line functions [PATCH 3/3] drm/vkms: Add support for XRGB2101010 [PATCH 0/3] drm/vkms: Miscelanious clarifications ---- this is the current series [PATCH 1/3] drm/vkms: Formatting and typo fix [PATCH 2/3] drm/vkms: Rename index to possible_crtc [PATCH 3/3] drm/vkms: Add documentation Magic: Preparing a sparse worktree Unable to cleanly apply series, see failure log below --- Application de drm/vkms: Code formatting Application de drm/vkms: Use drm_frame directly Application de drm/vkms: write/update the documentation for pixel conversion and pixel write functions [...] Application de drm/vkms: Add support for XRGB2101010 Application de drm/vkms: Miscelanious clarifications l'application de la rustine a échoué à 0026 drm/vkms: Miscelanious clarifications Quand vous avez résolu ce problème, lancez "git am --continue". Si vous préférez plutôt sauter ce patch, lancez "git am --skip". Pour restaurer la branche originale et arrêter de patcher, lancez "git am --abort". erreur : Pas de rustine valide sur l'entrée (permis avec "--allow-empty") astuce: Utilisez 'git am --show-current-patch=diff' pour visualiser le patch en échec --- Not fetching into FETCH_HEAD --- ● message-id: <20240516-yuv-v8-0-cf8d6f86430e@bootlin.com> ● message-id: <20240516-b4-new-color-formats-v1-0-74cf9fe07317@bootlin.com> ● message-id: <20240516-writeback_line_by_line-v1-0-7b2e3bf9f1c9@bootlin.com> ● base-commit: drm-misc/drm-misc-next - Could not cleanly apply patches If I understood correctly the code, the cover is removed at line 1408, but re-added at 1545 if the series contains multiple patches: if len(patches) == 1: mixin_cover(cbody, patches) else: add_cover(csubject, msgid_tpt, patches, cbody, seriests, thread=thread)
On Tue, May 21, 2024 at 05:42:10PM GMT, Louis Chauvet wrote: > If I understood correctly the code, the cover is removed at line 1408, > but re-added at 1545 if the series contains multiple patches: Yes, you're right -- my test case only had a single patch. Thanks for catching this. -K
diff --git a/src/b4/ez.py b/src/b4/ez.py index 76121ab02d03..5e6ed7f90900 100644 --- a/src/b4/ez.py +++ b/src/b4/ez.py @@ -919,7 +919,11 @@ def check_deps(cmdargs: argparse.Namespace) -> None: elif allgood: logger.info('Testing if all patches can be applied to %s', base_commit) tos, ccs, tstr, mypatches = get_prep_branch_as_patches(thread=False, movefrom=False, addtracking=False) - prereq_patches += [x[1] for x in mypatches] + if get_cover_strategy() == "commit": + # If the cover letter is stored as a commit, skip it to avoid empty patches + prereq_patches += [x[1] for x in mypatches[1:]] + else: + prereq_patches += [x[1] for x in mypatches] gitdir = os.getcwd() topdir = b4.git_get_toplevel(gitdir) if b4.git_commit_exists(topdir, base_commit): @@ -2451,7 +2455,9 @@ def get_info(usebranch: str) -> Dict[str, str]: info['preflight-checks-failing'] = bool(info['needs-editing'] or info['needs-auto-to-cc'] or info['needs-checking'] or info['needs-checking-deps']) - # Add commits information + # Add informations about the commits in this series + # `commit-<hash>`: stores the subject of each commit + # `series-<rev>`: stores the commit range for a particular revision for line in oneline: short, subject = line.split(maxsplit=1) info[f'commit-{short}'] = subject
When using the commit strategy for the cover letter, the first commit is empty. This commit can't be applied properly with git am, so just skip it. Signed-off-by: Louis Chauvet <louis.chauvet@bootlin.com> --- I tested to use --empty=drop/keep, but the error was the same, maybe because there are many patches to apply at the same time? --- src/b4/ez.py | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)