openstack/reviewday
Code Issues Proposed changes

LP specification name caching.

Browse Source 
This updates our LP calls to obtain blueprint information so that
they run faster. Previously calling spec.name would cause an LP hit
for each spec. Given that we call lp.specification many times during
a reviewday run this was causing slow runtimes.

What we do now is just scrap the blueprint name off of the Spec URL
(thus avoiding many of the extra LP hits).

Change-Id: I4f2cbcc262973b02e3a13b9bba1edde53d5e6d9e
This commit is contained in:
Dan Prince 2013-07-17 09:47:02 -04:00
parent e95925a075
commit 82c28bd48b
1 changed files with 8 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
reviewday/launchpad.py
View File

@ -18,11 +18,16 @@ class LaunchPad(object):
def specifications(self, project):
if project not in self.spec_cache:
specs = self.project(project).valid_specifications
self.spec_cache[project] = specs
cache = {}
for spec in specs:
spec_str = str(spec)
spec_name = spec_str[spec_str.index('+spec/') + 6:]
cache[spec_name] = spec
self.spec_cache[project] = cache
return self.spec_cache[project]
def specification(self, project, spec_name):
specs = self.specifications(project)
for spec in specs:
if spec.name == spec_name:
for name, spec in specs.iteritems():
if name == spec_name:
return spec