| |
@@ -43,16 +43,32 @@
|
| |
.filter(models.Build.package_id.in_(pkg_ids)) \
|
| |
.with_entities(func.max(models.Build.id)) \
|
| |
.group_by(models.Build.package_id)
|
| |
+
|
| |
+ # map package.id => package object in packages array
|
| |
packages_map = {package.id: package for package in packages}
|
| |
- builds = (
|
| |
- models.Build.query.filter(models.Build.id.in_(builds_ids))
|
| |
- .options(selectinload('build_chroots')))
|
| |
+
|
| |
+ builds = (models.Build.query.filter(models.Build.id.in_(builds_ids))
|
| |
+ .options(selectinload('build_chroots'))
|
| |
+ .yield_per(1000))
|
| |
+
|
| |
for build in builds:
|
| |
+ class SmallBuild():
|
| |
+ pass
|
| |
+
|
| |
if not build.package_id:
|
| |
continue
|
| |
- packages_map[build.package_id].latest_build = build
|
| |
|
| |
- return list(packages)
|
| |
+ small_build_object = SmallBuild()
|
| |
+ for param in ['state', 'status', 'pkg_version',
|
| |
+ 'submitted_on']:
|
| |
+ # we don't want to keep all the attributes here in memory, and
|
| |
+ # also we don't need any further info about assigned
|
| |
+ # build_chroot(s). So we only pick the info we need, and throw
|
| |
+ # the expensive objects away.
|
| |
+ setattr(small_build_object, param, getattr(build, param))
|
| |
+ packages_map[build.package_id].latest_build = small_build_object
|
| |
+
|
| |
+ return packages
|
| |
|
| |
@classmethod
|
| |
def get_list_by_copr(cls, copr_id, package_name):
|
| |
praiskup
commented 4 years ago
To have properly generated "stream" feeling from streamed jinja, we need
to have streamed sqlalchemy result set as well. So yield_per helped a
lot in case of /builds/.
With packages, it's not that trivial. But the set of returned packages
is much smaller than set of returned builds (and build_chroots!). So
let's only return we are interested in (package + latest_build status,
and some small things).