summaryrefslogtreecommitdiff
path: root/coverage/sqldata.py
Commit message (Collapse)AuthorAgeFilesLines
...
* Meta data stored as key/valueNed Batchelder2019-08-061-14/+22
|
* Foreign keysNed Batchelder2019-08-061-2/+8
|
* A little more discipline for blob convertersNed Batchelder2019-07-311-5/+12
|
* numbits is the new name for the binary line numbersNed Batchelder2019-07-311-18/+27
|
* Refactor numbits into their own filesNed Batchelder2019-07-311-33/+7
|
* Don't try to delete a file called ':memory:'Ned Batchelder2019-07-201-0/+2
|
* z-compressed dumps and loadsNed Batchelder2019-07-191-12/+17
|
* Bitmaps for line numbers, 10% of the size.Ned Batchelder2019-07-151-32/+94
|
* :memory: supportNed Batchelder2019-07-151-6/+15
|
* CoverageData.dumps and loads for serializationNed Batchelder2019-07-151-8/+21
|
* Merge branch 'master' into isolate_osNed Batchelder2019-07-101-6/+68
|\
| * Remove the JSON data codeNed Batchelder2019-07-101-6/+68
| |
* | Isolate the os module to protect from aggressive mocking interfearing with ↵Matt Bachmann2019-07-091-1/+2
|/ | | | database operations
* SQLite will keep comments that are inside, not outsideNed Batchelder2019-07-081-7/+7
| | | | | This way, ".schema" in the SQLite prompt will show the comments for the tables.
* Merge branch 'master' into Fix-typoNed Batchelder2019-07-071-8/+24
|\
| * Avoid useless or redundant db operations. Faster.Ned Batchelder2019-07-071-3/+17
| | | | | | | | | | | | Moving operations into the "with self._connect" means less opening and closing of the database. Returning early if there is no data to write avoids writing empty contexts.
| * Log connections properlyNed Batchelder2019-07-071-2/+2
| |
| * Clarify SqliteDbNed Batchelder2019-07-071-3/+5
| |
* | Fix typoMin ho Kim2019-07-081-2/+2
|/
* Clean up and test filtering contexts for reportingNed Batchelder2019-07-011-9/+8
|
* Add comments to the SQL schema, and keep them when creating the dbNed Batchelder2019-06-111-22/+28
|
* Fix two pylint warningsNed Batchelder2019-06-111-2/+3
|
* Make --contexts available for HTML reportsNed Batchelder2019-06-101-12/+21
|
* Tricky case: Thread switching is getting in the way.Stephan Richter2019-06-101-1/+6
|
* Add ability to report contexts for all lines of a file.Stephan Richter2019-06-101-0/+34
|
* Sigh, sqlite3 does not support proper array serialization.Stephan Richter2019-06-101-8/+10
|
* Initialize attributes.Stephan Richter2019-06-101-0/+2
|
* Support for multiple, glob-based contexts in result queries.Stephan Richter2019-06-101-8/+33
|
* Don't name new method the same as old attributeNed Batchelder2019-04-201-1/+1
|
* Fix #708: .filename is private, and don't combine over ourselvesNed Batchelder2019-04-151-13/+21
|
* Oops, clean up some unused listcompsNed Batchelder2019-04-091-7/+6
|
* A little clean-up from pr #760Ned Batchelder2019-04-091-9/+2
|
* Merge pull request #723 from blueyed/fix-702Ned Batchelder2019-04-081-2/+2
|\ | | | | CoverageSqliteData._file_id: use "insert or replace"
| * CoverageSqliteData._file_id: fall back to SELECTDaniel Hahler2019-03-311-2/+2
| | | | | | | | Fixes https://github.com/nedbat/coveragepy/issues/702
* | Merge pull request #760 from Shoobx/fix-coverage-with-threadsNed Batchelder2019-04-081-17/+25
|\ \ | | | | | | Ensure sqldata thread safety.
| * | Make thread code Py2 compatible.Stephan Richter2019-01-261-8/+14
| | |
| * | Use one conenction per thread to avoid tricky race conditions. (The ref ↵Stephan Richter2019-01-251-17/+19
| | | | | | | | | | | | counter is simply not good enough.)
* | | Fix formatting, and credit StephanNed Batchelder2019-04-071-31/+37
| | |
* | | Merge pull request #765 from Shoobx/fast-sqldata-updateNed Batchelder2019-04-071-45/+122
|\ \ \ | |_|/ |/| | Fast CoverageSqliteData Update/Combine
| * | Some cleanup (left from first versions) and localized file path lookup which ↵Stephan Richter2019-01-271-18/+11
| | | | | | | | | | | | takes off another 20%.
| * | Use efficient database queries to combine data. This algorithm is 25-30x ↵Stephan Richter2019-01-261-45/+129
| |/ | | | | | | | | | | faster than the original one. Given that context-included coverage has about 50x the data of the old format (on our code base), this means that combining is now only 2-3x slower than before.
* | Make sure that arcs() and lines() only returns distinct line combination ↵Stephan Richter2019-02-171-2/+2
|/ | | | values and not one entry for each context.
* Use one transaction to speed combiningNed Batchelder2018-10-151-39/+45
|
* Defer using the database when calling set_context #716Ned Batchelder2018-10-141-11/+14
| | | | | | | | | | | The collector calls set_context() before any code is run. If we touch the database there, it will get created *very* early. This causes problems with pytest-cov, which will delete those early-created files when erasing data. By deferring the database access until add_lines is called, the data file stays off the disk until the collection is done (or until the context switches), which avoids the problem.
* Fewer conditionals for debug outputNed Batchelder2018-10-131-11/+11
|
* Debugging improvementsNed Batchelder2018-10-131-2/+2
|
* Record the sys.argv in the dbNed Batchelder2018-09-281-3/+5
|
* Faster combiningNed Batchelder2018-09-271-7/+13
|
* CoverageSqliteData.__nonzero__: do not create DBDaniel Hahler2018-09-241-0/+2
| | | | This makes is more lazy and avoids creating an empty DB unnecessarily.
* Dynamic contextsNed Batchelder2018-09-231-0/+2
|
View Lorry Controller Status