Граф коммитов

1419 Коммитов

Автор SHA1 Сообщение Дата
Marco Castelluccio 3324f91f00 Keep using standard json library in the test scheduling history retrieval task
The orjson library doesn't support stream reading from a file, so here
the standard json library is preferable.
2019-10-18 01:25:31 +01:00
dependabot-preview[bot] 25e29d4fe0 Bump rs-parsepatch from 0.2.3 to 0.2.4 (#1024)
Bumps [rs-parsepatch]() from 0.2.3 to 0.2.4.

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-17 11:27:47 +02:00
Marco Castelluccio b4a56e7dcd Version 0.0.128 2019-10-16 16:28:06 +02:00
Marco Castelluccio 517519d7d6 Add a test for the removal of a non-existent DB 2019-10-16 16:20:53 +02:00
Marco Castelluccio 72e8961675 Don't fail when trying to remove a DB which does not exist
For example, the bug retriever is trying to delete old bugs, but
the deletion fails when we increase the version of the bugs DB and
we need to download a completely new one.
2019-10-16 16:19:11 +02:00
Marco Castelluccio ef767034df orjson.dumps returns bytes and not str, so we don't need no encoding 2019-10-16 16:18:50 +02:00
Marco Castelluccio 89e88981b7 Use orjson in cases where we need high performance 2019-10-16 16:10:29 +02:00
Marco Castelluccio deed153210 Version 0.0.127 2019-10-16 15:18:36 +02:00
dependabot-preview[bot] 90c676a4ce Bump taskcluster from 19.0.0 to 20.0.0 (#1020)
Bumps [taskcluster](https://github.com/taskcluster/taskcluster) from 19.0.0 to 20.0.0.
- [Release notes](https://github.com/taskcluster/taskcluster/releases)
- [Changelog](https://github.com/taskcluster/taskcluster/blob/master/CHANGELOG.md)
- [Commits](https://github.com/taskcluster/taskcluster/compare/v19.0.0...v20.0.0)

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-16 14:47:04 +02:00
dependabot-preview[bot] 94d6d5cb4c Bump rs-parsepatch from 0.2.2 to 0.2.3 (#1022)
Bumps [rs-parsepatch]() from 0.2.2 to 0.2.3.

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-16 14:36:49 +02:00
Marco Castelluccio b2ab81d3e8 Pin CI image to Python 3.7 2019-10-16 14:07:03 +02:00
Marco Castelluccio 898dea8bdb Count total number of commits while iterating to get the file counts 2019-10-16 10:09:43 +02:00
Marco Castelluccio 7007118e88 Update Bugzilla DB version, as there are some component changes 2019-10-16 10:07:30 +02:00
Marco Castelluccio b3a3763a86 Version 0.0.126 2019-10-15 23:25:26 +02:00
Marco Castelluccio e230508a30 Exctractor's transform argument must be a function returning an iterable
Regression from c9f8613d79
2019-10-15 23:04:44 +02:00
Marco Castelluccio a6a4b471dd Version 0.0.125 2019-10-15 14:20:26 +02:00
Marco Castelluccio 3d36786b58 Install libffi-dev, required for cffi 2019-10-15 14:20:09 +02:00
Marco Castelluccio 91ea555aca Version 0.0.124 2019-10-15 12:09:51 +02:00
Marco Castelluccio 7f8e08c20d Add a task to train the test selection model 2019-10-12 17:31:28 +01:00
Marco Castelluccio c16bc7a362 Rename test_select.py to testselect.py
To avoid issues with the way the trainer script stores models on disk.
2019-10-12 17:31:03 +01:00
Marco Castelluccio 795acbcbc9 Specify DBs required by the test selection model 2019-10-12 17:30:50 +01:00
Marco Castelluccio 22a26e0873 Define required DBs in a attribute of the model
This way the trainer script doesn't have to rely on a brittle
check on the base class of the model

Fixes #520
2019-10-12 16:10:46 +01:00
Marco Castelluccio 863c1fbc48 Implement a WIP model for test selection 2019-10-12 00:58:39 +01:00
Marco Castelluccio 1bf5a8edf3 Make train/test split configurable by subclasses of Model
Some models might want to implement a different way to split between
train and test sets. For example, for test scheduling we'll need to
use the past past as the training set, and the recent past as the
test set. We can't mix them or the evaluation will be falsified.
2019-10-12 00:58:39 +01:00
Marco Castelluccio c68d469489 Make it possible to use the test job related features with the commit extractor 2019-10-12 00:58:39 +01:00
Marco Castelluccio ef0e4baa16 Add a module to extract test job related features 2019-10-12 00:58:39 +01:00
Marco Castelluccio c9f8613d79 During training, don't build a list with all trainable elements, but use a generator
It could use too much memory if there are too many elements.
2019-10-12 00:58:39 +01:00
Marco Castelluccio 10bb0982ab Don't pop 500 elements if we have less than that in the list 2019-10-10 20:51:12 +01:00
Marco Castelluccio 92062a0b83 Version 0.0.123 2019-10-10 18:52:27 +01:00
Marco Castelluccio 2cfd8fc01a Try using relman-svc-compute for the test scheduling history retrieval task 2019-10-10 18:52:08 +01:00
Marco Castelluccio 71342d12c9 Actually analyze 500 commits at a time, ignoring skipped commits
Otherwise, we risk never ending loop of tasks, as if we skip 500 we never get out.
2019-10-10 18:50:51 +01:00
dependabot-preview[bot] fcda1b23e4 Bump microannotate from 0.0.15 to 0.0.16 (#1016)
Bumps [microannotate]() from 0.0.15 to 0.0.16.

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-10 16:03:30 +01:00
Marco Castelluccio f6518f6375 Version 0.0.122 2019-10-10 11:39:12 +01:00
Marco Castelluccio 6ace4d78bc Use relman-svc-memory for the test scheduling history retriever task 2019-10-10 11:08:45 +01:00
dependabot-preview[bot] 3e86909d90 Bump jsonschema from 3.0.2 to 3.1.1 (#1014)
Bumps [jsonschema](https://github.com/Julian/jsonschema) from 3.0.2 to 3.1.1.
- [Release notes](https://github.com/Julian/jsonschema/releases)
- [Changelog](https://github.com/Julian/jsonschema/blob/master/CHANGELOG.rst)
- [Commits](https://github.com/Julian/jsonschema/compare/v3.0.2...v3.1.1)

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-10 00:54:53 +01:00
Marco Castelluccio 01349e8467 Version 0.0.121 2019-10-09 15:03:34 +01:00
Marco Castelluccio 0fefacba0b Get past task failures with specific files, directories, components, file types 2019-10-09 15:03:07 +01:00
Marco Castelluccio d0cd6a960f Don't write useless whitespaces in adr configuration 2019-10-09 15:03:07 +01:00
Bastien Abadie 7a52fd77e6
Support utf-8 mercurial message & author. (#1012) 2019-10-08 23:02:13 +02:00
dependabot-preview[bot] 86823f36d7 Bump taskcluster from 18.0.3 to 19.0.0 (#1011)
Bumps [taskcluster](https://github.com/taskcluster/taskcluster) from 18.0.3 to 19.0.0.
- [Release notes](https://github.com/taskcluster/taskcluster/releases)
- [Changelog](https://github.com/taskcluster/taskcluster/blob/master/CHANGELOG.md)
- [Commits](https://github.com/taskcluster/taskcluster/compare/v18.0.3...v19.0.0)

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-07 23:45:17 +01:00
Marco Castelluccio f5199b9655 Version 0.0.120 2019-10-07 21:13:54 +01:00
Marco Castelluccio 31fe020a42 Make bug_classifier script output importance plot 2019-10-07 19:15:27 +01:00
Marco Castelluccio 49a0e61c4a Remove feature_names parameter from print_feature_importances call, as it was removed 2019-10-07 19:14:34 +01:00
Marco Castelluccio 2f40304242 Don't fail if MethodDefectPredictor did not generate a result file
It happens when there are no methods to analyze.
2019-10-07 16:40:11 +01:00
Marco Castelluccio 88fbad5140 Version 0.0.119 2019-10-07 15:43:39 +01:00
Marco Castelluccio 01c14b85f1 Update MethodDefectPredictor to have riskiness confidence for methods and a bug fix when no methods are analyzed 2019-10-07 15:29:56 +01:00
Marco Castelluccio b42662367d Add more regression/bug/nobug labels 2019-10-07 13:32:24 +01:00
Marco 882d22f640
Add a few more file types (#1010) 2019-10-07 12:23:18 +01:00
dependabot-preview[bot] f6a6040b13 Bump pytest from 5.2.0 to 5.2.1 (#1009)
Bumps [pytest](https://github.com/pytest-dev/pytest) from 5.2.0 to 5.2.1.
- [Release notes](https://github.com/pytest-dev/pytest/releases)
- [Changelog](https://github.com/pytest-dev/pytest/blob/master/CHANGELOG.rst)
- [Commits](https://github.com/pytest-dev/pytest/compare/5.2.0...5.2.1)

Signed-off-by: dependabot-preview[bot] <support@dependabot.com>
2019-10-06 18:31:48 +01:00
Marco Castelluccio 9efbdff8f6 Use tf.keras and stop using keras 2019-10-05 21:54:26 +01:00