Commit graph

5727 commits

Author SHA1 Message Date
09ebbc9cc6
findDocuments: improve performance of positions queries
References #2935
2016-06-28 16:36:55 +01:00
8d25e3be78
findDocuments: improve entity query performance
When I implemented this in 9a4c24c, there were not many rows in
entity_documentproperties in the database here. Now that there are,
computing the document_document -> entity_documentproperties ->
entity_entity join and then filtering is really, really slow. Postgres
seems to materialize the whole join and then scan it.

If we get a set of matching document IDs for the entity query in a
subquery, and then just filter with IN on that, things are much faster:
scan entity_entity; in a nested loop, get the document_ids via
entity_documentproperties; hash this set; and then scan
document_document.

Searching for a single character, this brings the query from ~1.1s to
~400ms. Searching for a full word, ~800ms to 120ms

This condition is getting really ugly -- I am sorry!

References #2935
2016-06-28 16:33:01 +01:00
92a4e77f9b
chunkupload: typo: uplaod -> upload 2016-06-28 16:14:04 +01:00
j
5aeffcfb6a check first audio track 2016-06-27 16:51:18 +02:00
j
adfcc1cb27 never set display aspect ratio to 0:0 2016-06-27 16:08:30 +02:00
j
8ac78f3bd6 remove unused force flag from make_poster, update_timeline 2016-06-26 23:24:11 +02:00
j
0f9e80e1e6 avoid saving item twice 2016-06-26 23:22:27 +02:00
j
de9b062d63 make sure existing index is using gin 2016-06-26 16:55:58 +02:00
j
ab0dfddf31 set SECURE_PROXY_SSL_HEADER by default 2016-06-26 15:34:19 +02:00
j
0d89ad640b ignore some broken audio codecs 2016-06-26 15:33:52 +02:00
j
92f642cbac pcm sound can have no codec 2016-06-26 14:41:58 +02:00
j
2cec1b9ad5 s/import Image/from PIL import Image/g 2016-06-25 20:39:29 +02:00
j
4785f314cb Add VP9/Opus support, use VP8 by default
- support vp9 and opus
- switch to 2 pass encoding
- use ffmpeg -movflags +faststart instead of qtfaststart
2016-06-23 17:36:41 +02:00
j
aaacc48259 only save if update_external fails 2016-06-20 18:28:05 +02:00
j
d83647c4a5 don't hide oxtimelines errors 2016-06-20 18:27:31 +02:00
j
6dcbcdd19c dont update timeline in update_selected, remove unused async get_item case 2016-06-16 14:48:54 +02:00
j
0486d62ec9 use absolute path 2016-06-16 14:48:09 +02:00
j
f25218466b formating 2016-06-16 14:48:01 +02:00
j
70f34bfde9 typo 2016-06-15 19:13:00 +02:00
j
e3c5ab18c7 only update itemsort if name is changed 2016-06-15 18:31:40 +02:00
j
22f83288c5 avoid looking up item twice 2016-06-15 18:29:09 +02:00
j
7c53dca65b less async item creation 2016-06-15 18:12:59 +02:00
j
b2a9a5f711 space 2016-06-15 17:56:31 +02:00
j
3c1f4a8c95 dont call module 2016-06-15 17:55:57 +02:00
j
b010aca0a9 s/taskId/id/ 2016-06-15 15:45:51 +02:00
j
a0fc6ffadc typo 2016-06-15 14:55:45 +02:00
j
f4cbe6a114 return empty sequences if no data timeline exists 2016-06-15 14:48:02 +02:00
j
af0e0cffe8 person can be removed again, let async itemsort fail without exception 2016-06-15 14:34:46 +02:00
j
fd9d3bdabf flake8 + map->[] 2016-06-15 14:34:46 +02:00
j
93734a4fbc lxc install cleanup 2016-06-15 14:34:46 +02:00
j
542fe8a860 update symlinks too, follow up to ed21517 2016-06-13 10:30:22 +01:00
j
437c19f64c ignore share 2016-06-09 12:57:38 +02:00
j
cb6203ff82 use same folder structure in etc/systemd as used once installed 2016-06-09 12:50:51 +02:00
j
d7a9ca28cb also create /var/log/pandora 2016-06-09 12:48:42 +02:00
j
5e3b4b9c2b change port in gunicorn_config.py 2016-06-09 12:42:00 +02:00
j
ed215174f8 install systemd services in /etc/systemd/system, /lib/systemd could be mounted read only in the future 2016-06-09 12:38:13 +02:00
j
05c4cfcbc8 add space and other flake8 cleanups 2016-05-28 11:30:43 +02:00
j
5e149a5cb8 add space and other flake8 cleanups 2016-05-28 11:26:46 +02:00
j
225259e521 add space and other flake8 cleanups 2016-05-28 11:18:51 +02:00
j
f21e8413fb use get_random_string 2016-05-28 11:18:51 +02:00
j
7fdaf6d1ce include Access-Control-Allow-Origin in 404 not found response 2016-05-27 11:51:47 +02:00
j
d20620a716 name can contain &, decode html first. fixes #2917 2016-05-17 15:25:18 +00:00
05e6118a88
findAnnotations: include duration alongside result count
fixes #2921
2016-05-05 15:54:25 +01:00
j
41cc8e3573 expose encoding status via api 2016-05-05 10:49:34 +02:00
j
be163826ef Merge remote-tracking branch 'wjt/fix-migrations' 2016-05-05 10:48:24 +02:00
39b9b48be2
archive: fix migrations for upload_to function renamings
9c75526 renamed these functions. The function doesn't affect the DB
schema so it should be safe to just modify the migraiton.
2016-05-04 17:01:44 +01:00
e29ea230fb
Add migration for Document.documentproperties ref
This should have been included with a8dcbbb, which changed the
related_name to access DocumentProperties from Document. (There's no
actual change to the database.)
2016-05-04 16:55:11 +01:00
j
0f28a2b7d5 fix queue status 2016-04-30 14:15:13 +02:00
j
9c7552699f fix upload_to callbacks 2016-04-29 13:46:55 +02:00
2812834ce3
findAnnotations: don't lowercase ids (fixes #2916)
Without this fix, a condition like:

     {key: 'id', operator: '==', value: 'A/B'}

gets mapped to:

     public_id__exact=('A/B'.lower())

which is wrong.

I introduced this bug in b3df5b8. I didn't catch it because I was
mostly interested in the 'layer' key -- but layer names are
conventionally lowercase anyway so lowercasing them had no effect.
2016-04-29 11:03:45 +01:00