def _load_characteristics(ret, repo):
if ret is None:
ret = dict()
missing = all_messages_in_time_window - ret.keys()
if len(missing) == 0:
return ret, False
missing = load_linux_mail_characteristics(repo, missing,
maintainers_version,
clustering)
return {**ret, **missing}, True
def evaluate_patches(config, prog, argv):
if config.mode != config.Mode.MBOX:
log.error('Only works in Mbox mode!')
return -1
if config.mbox_use_patchwork_id:
log.error('pasta evaluate_patches does not work with '
'USE_PATCHWORK_ID = true')
repo = config.repo
_, clustering = config.load_cluster()
clustering.optimize()
config.load_ccache_mbox()
repo.mbox.load_threads()
patches = set()
upstream = set()
for d, u in clustering.iter_split():
patches |= d
upstream |= u
all_messages_in_time_window = repo.mbox.get_ids(config.mbox_time_window,
allow_invalid=True)
tags = {x[0] for x in repo.tags if not x[0].startswith('v2.6')}
tags |= {x[0] for x in repo.tags if x[0].startswith('v2.6.39')}
maintainers_version = load_maintainers(config, tags)
characteristics = \
load_linux_mail_characteristics(config, maintainers_version, clustering,
all_messages_in_time_window)
relevant = get_relevant_patches(characteristics)
log.info('Identify ignored patches...')
ignored_patches, ignored_patches_related = get_ignored(
characteristics, clustering, relevant)
dump_characteristics(repo, characteristics, ignored_patches_related,
relevant, config.f_characteristics)
call(['./R/ignored_patches.R', config.d_rout, config.f_characteristics])
def prepare_ignored_patches(config, clustering):
def _get_kv_rc(linux_version):
tag = linux_version.split('-rc')
kv = tag[0]
rc = 0
if len(tag) == 2:
rc = int(tag[1])
return kv, rc
repo = config.repo
repo.mbox.load_threads()
patches = set()
upstream = set()
for d, u in clustering.iter_split():
patches |= d
upstream |= u
all_messages_in_time_window = repo.mbox.get_ids(config.mbox_time_window,
allow_invalid=True)
tags = {x[0] for x in repo.tags if not x[0].startswith('v2.6')}
tags |= {x[0] for x in repo.tags if x[0].startswith('v2.6.39')}
maintainers_version = load_maintainers(config, tags)
characteristics = \
load_linux_mail_characteristics(config, maintainers_version, clustering,
all_messages_in_time_window)
relevant = get_relevant_patches(characteristics)
log.info('Identify ignored patches...')
# Calculate ignored patches
ignored_patches = {
patch
for patch in relevant if not characteristics[patch].is_upstream
and not characteristics[patch].has_foreign_response
}
# Calculate ignored patches wrt to other patches in the cluster: A patch is
# considered as ignored, if all related patches were ignoreed as well
ignored_patches_related = \
{patch for patch in ignored_patches if False not in
[characteristics[x].has_foreign_response == False
for x in (clustering.get_downstream(patch) & relevant)]}
num_relevant = len(relevant)
num_ignored_patches = len(ignored_patches)
num_ignored_patches_related = len(ignored_patches_related)
log.info('Found %u ignored patches' % num_ignored_patches)
log.info('Fraction of ignored patches: %0.3f' %
(num_ignored_patches / num_relevant))
log.info('Found %u ignored patches (related)' %
num_ignored_patches_related)
log.info('Fraction of ignored related patches: %0.3f' %
(num_ignored_patches_related / num_relevant))
log.info('Dumping characteristics...')
ignored_target = ignored_patches_related
# Alternative analysis:
#ignored_target = ignored_patches
with open(config.f_characteristics, 'w') as csv_file:
csv_fields = [
'id', 'from', 'list', 'list_matches_patch', 'kv', 'rc', 'ignored',
'time'
]
writer = csv.DictWriter(csv_file, fieldnames=csv_fields)
writer.writeheader()
for message_id in sorted(relevant):
c = characteristics[message_id]
kv, rc = _get_kv_rc(c.linux_version)
mail_from = c.mail_from[1]
for list in repo.mbox.get_lists(message_id):
list_matches_patch = False
for subsys in c.maintainers.values():
lists = subsys[0]
if list in lists:
list_matches_patch = True
break
row = {
'id': message_id,
'from': mail_from,
'list': list,
'list_matches_patch': list_matches_patch,
'kv': kv,
'rc': rc,
'ignored': message_id in ignored_target,
'time': c.date,
}
writer.writerow(row)
log.info('Calling R...')
call([
'./analyses/ignored_patches.R', config.d_rout, config.f_characteristics
])