hub/databricks-sql-connector
District: requests · urllib3databricks/databricks-sql-python
- Bump to version 4.4.99
#235opened by databricks-sql-connector-bot· merged session 09-04 11:34
1 - perf: hand out a SEA JSON row without a call per row (1.07-1.10x on sea_fetchone_json)1
#234opened by databricks-sql-connector-bot· merged session 09-04 11:34
- Bump to version 4.4.98
#233opened by databricks-sql-connector-bot· merged session 09-04 11:03
0 - perf: decide a SEA fetchone's branch per block, not per row (1.25-1.43x on sea_fetchone_arrow_*)
#232opened by databricks-sql-connector-bot· merged session 09-04 11:03
1 - Bump to version 4.4.97
#231opened by databricks-sql-connector-bot· merged session 09-04 10:32
0 - perf: parse a target host's origin once instead of once per request (3.4-4.1x on the CloudFetch proxy decision, 1.27-1.51x on a proxied request)
#230opened by databricks-sql-connector-bot· merged session 09-04 09:57
1 - ci: run the unit tests against the urllib3 floor we declare
#229opened by databricks-sql-connector-bot· merged session 09-04 09:27
1 - Bump to version 4.4.96
#228opened by databricks-sql-connector-bot· merged session 09-04 08:37
1 - perf: remember a proxied client per-host proxy decision (2.2-3.5x on the decision, 1.25-1.51x on a proxied request)1
#227opened by databricks-sql-connector-bot· merged session 09-04 08:37
- Bump to version 4.4.95
#225opened by databricks-sql-connector-bot· merged session 09-04 07:48
1 - perf: let a column reader take its type class as already decided (1.08-1.18x on a one-row statement)1
#224opened by databricks-sql-connector-bot· merged session 09-04 07:48
- bench: census the pool keys the two PoolManager shapes derive
#222opened by databricks-sql-connector-bot· merged session 09-04 07:21
1 - Bump to version 4.4.940
#221opened by databricks-sql-connector-bot· merged session 09-04 06:46
- perf: decide an Arrow column reader once per type, not once per column (1.14-1.22x on a one-row statement)1
#220opened by databricks-sql-connector-bot· merged session 09-04 06:46
- Bump to version 4.4.93
#219opened by databricks-sql-connector-bot· merged session 09-04 06:27
0 - perf: share one row class between result sets over the same columns (1.8-2.0x on a one-row statement)
#218opened by databricks-sql-connector-bot· merged session 09-04 05:36
1 - Bump to version 4.4.920
#217opened by databricks-sql-connector-bot· merged session 09-04 03:37
- perf: give SEA ARROW_STREAM fetchone() the growing block the other backends have (29-46x on sea_fetchone_arrow_*)2
#216opened by databricks-sql-connector-bot· merged session 09-04 03:37
- Bump to version 4.4.91
#215opened by databricks-sql-connector-bot· merged session 09-04 03:13
0 - perf: grow the kernel fetchone() block too, instead of fixing it at 256 rows (1.44-1.70x on kernel_fetchone_*)
#214opened by databricks-sql-connector-bot· merged session 09-04 03:13
1 - Bump to version 4.4.90
#213opened by databricks-sql-connector-bot· merged session 09-04 02:31
1 - perf: let the row buffer's length be the position fetchone reads from (1.14x on fetchone_*, 1.16x on `for row in result_set`)1
#212opened by databricks-sql-connector-bot· merged session 09-04 02:31
- bench: give the columnar cell the tuple columns the backend builds (fetchall_columnar was 25-30% fixture)
#210opened by databricks-sql-connector-bot· merged session 09-04 02:10
1 - Bump to version 4.4.89
#209opened by databricks-sql-connector-bot· merged session 09-04 01:46
0 - perf: look up a Thrift command type instead of rebuilding the map per RPC (1.55-1.59x on make_request's wrapper)
#208opened by databricks-sql-connector-bot· merged session 09-04 01:46
1 - Bump to version 4.4.88
#207opened by databricks-sql-connector-bot· merged session 09-04 01:19
0 - perf: filter SEA table types without deciding case-sensitivity per row (1.90x-1.96x on the filter)
#206opened by databricks-sql-connector-bot· merged session 09-04 01:19
1 - Bump to version 4.4.87
#205opened by databricks-sql-connector-bot· merged session 09-04 00:52
0 - perf: read only the state when polling a SEA statement's state (1.12-1.27x on a poll)
#204opened by databricks-sql-connector-bot· merged session 09-04 00:52
2 - Bump to version 4.4.86
#203opened by databricks-sql-connector-bot· merged session 09-04 00:05
1 - perf: localise a pytz timestamp column through Python ints, and once when it spans no transition (1.20-1.36x on the column read)1
#202opened by databricks-sql-connector-bot· merged session 09-04 00:05
- bench: price the column -> arrow decode, which turns out to have no caller (and build its type table once)1
#200opened by databricks-sql-connector-bot· merged session 09-03 23:29
- Bump to version 4.4.85
#199opened by databricks-sql-connector-bot· merged session 09-03 23:16
0 - perf: keep the thrift read buffer armed across arrow batches under 512 KiB (1.13-1.77x on narrow-batch arrow reads)
#198opened by databricks-sql-connector-bot· merged session 09-03 22:35
1 - Bump to version 4.4.84
#197opened by databricks-sql-connector-bot· merged session 09-03 21:28
0 - perf: build a SEA CloudFetch link's Thrift struct positionally (1.35x on the conversion, 1.24x on a link batch)
#196opened by databricks-sql-connector-bot· merged session 09-03 21:28
2 - bench: price a percent escape in a presigned URL's path (+2.3 to +7.7 us on urllib3 2.7.60, free on 2.7.61)1
#195opened by databricks-sql-connector-bot· merged session 09-03 20:02
- bench: count the URL machinery, and name the urllib3 that actually ran
#194opened by databricks-sql-connector-bot· merged session 09-03 19:01
1 - Bump to version 4.4.830
#193opened by databricks-sql-connector-bot· merged session 09-03 18:25
- perf: stop parsing the request URL per request when nothing reads the parse (1.10-1.12x on every CloudFetch chunk and staging PUT)1
#192opened by databricks-sql-connector-bot· merged session 09-03 18:25
- Bump to version 4.4.82
#191opened by databricks-sql-connector-bot· merged session 09-03 18:03
0 - perf: take CloudFetch links off the front of a deque, and ask the log level once per batch (2.6x on manager init, 1.5-1.8x on schedule)
#190opened by databricks-sql-connector-bot· merged session 09-03 18:03
1 - docs: give the --sea cells the two-arm run their own caveat asks for
#189opened by databricks-sql-connector-bot· merged session 09-03 17:38
1 - docs: the count table has two axes, and 4.4.81 only moved one of them
#188opened by databricks-sql-connector-bot· merged session 09-03 17:21
0 - Bump to version 4.4.81
#187opened by databricks-sql-connector-bot· merged session 09-03 17:21
0 - perf: hand a UnifiedHttpClient request to urllib3 through urlopen, not request (1.07x on a staging PUT, ~2 us per CloudFetch chunk)1
#186opened by databricks-sql-connector-bot· merged session 09-03 16:45
- bench: count_http_calls' UnifiedHttpClient shapes send the user agent a connection sends (iteritems/items were reading 0, they are 1 each)
#185opened by databricks-sql-connector-bot· merged session 09-03 16:24
0 - bench: count which urllib3 entry points a real request enters (counts, not times)1
#184opened by databricks-sql-connector-bot· merged session 09-03 13:50
- bench: bench_transport's SEA cells send the requests the connector sends (poll and close were understating by ~14.5%)1
#183opened by databricks-sql-connector-bot· merged session 09-03 12:59
- Bump to version 4.4.800
#182opened by databricks-sql-connector-bot· merged session 09-03 12:05
- perf: the once-per-response half of the SEA response parse (1.73x on the parse, 1.02-1.06x in path)1
#181opened by databricks-sql-connector-bot· merged session 09-03 12:05
- Bump to version 4.4.79
#180opened by databricks-sql-connector-bot· merged session 09-03 11:40
0 - perf: build SEA chunk and link objects positionally (1.24-1.32x on a 512-chunk poll, 1.78-1.95x on a link batch)
#179opened by databricks-sql-connector-bot· merged session 09-03 11:40
1 - Bump to version 4.4.78
#178opened by databricks-sql-connector-bot· merged session 09-03 11:12
0 - perf: log a CloudFetch link lazily, not through .format() (1.45-1.61x per link)
#177opened by databricks-sql-connector-bot· merged session 09-03 11:12
1 - Bump to version 4.4.770
#176opened by databricks-sql-connector-bot· merged session 09-03 10:22
- perf: retype only the decimal columns of an arrow table (2.8-4.1x on the decimal retype, 1.4-1.8x on a fetched batch)0
#175opened by databricks-sql-connector-bot· merged session 09-03 10:22
- test: state the result-format memo's properties against an empty memo (fixes #172)
#174opened by databricks-sql-connector-bot· merged session 09-03 09:55
0 - Bump to version 4.4.76
#173opened by databricks-sql-connector-bot· merged session 09-03 09:25
0 - perf: defer importlib.metadata out of the connect() import (1.07-1.09x on the connector's import)
#171opened by databricks-sql-connector-bot· merged session 09-03 08:49
4 - Bump to version 4.4.75
#170opened by databricks-sql-connector-bot· merged session 09-03 08:27
0 - perf: defer the SEA backend and dateutil out of the connect() import (1.09-1.17x on the connector's import)
#168opened by databricks-sql-connector-bot· merged session 09-03 07:52
1 - bench: the --flush cell is 2.3x cheaper on urllib3 2.7.51, and it is not ours
#167opened by databricks-sql-connector-bot· merged session 09-03 07:27
0 - Bump to version 4.4.740
#166opened by databricks-sql-connector-bot· merged session 09-03 06:43
- perf: stop re-polling for a state the async loop has just been told (one fewer round-trip per get_async_execution_result)0
#165opened by databricks-sql-connector-bot· merged session 09-03 06:43
- Bump to version 4.4.73
#164opened by databricks-sql-connector-bot· merged session 09-03 06:20
0 - perf: hand a SEA request to urllib3 through urlopen, not request (1.6x per statement submit)
#163opened by databricks-sql-connector-bot· merged session 09-03 05:44
5 - Bump to version 4.4.72
#162opened by databricks-sql-connector-bot· merged session 09-03 05:22
0 - perf: hand a thrift request to urllib3 through urlopen, not request (1.8-2.3x per flush)
#161opened by databricks-sql-connector-bot· merged session 09-03 04:47
1 - Bump to version 4.4.71
#160opened by databricks-sql-connector-bot· merged session 09-03 04:28
0 - perf: write a thrift request against a compacted spec, not 3354 empty slots (3-8x per RPC)
#159opened by databricks-sql-connector-bot· merged session 09-03 03:48
1 - Bump to version 4.4.700
#158opened by databricks-sql-connector-bot· merged session 09-03 03:12
- perf: read a CloudFetch link's expiry with fromisoformat, not dateutil (21-26x per link)0
#157opened by databricks-sql-connector-bot· merged session 09-03 03:12
- Bump to version 4.4.69
#156opened by databricks-sql-connector-bot· merged session 09-03 02:41
0 - perf: scan for the whole VARIANT marker, not just its key (8-10x on a schema that names other SQL types)
#155opened by databricks-sql-connector-bot· merged session 09-03 02:41
1 - Bump to version 4.4.680
#154opened by databricks-sql-connector-bot· merged session 09-03 01:52
- perf: skip the unsafe-logger record nobody is listening for (3.4-3.8x on `make_request`)0
#153opened by databricks-sql-connector-bot· merged session 09-03 01:52
- Bump to version 4.4.67
#152opened by databricks-sql-connector-bot· merged session 09-03 01:40
0 - perf: skip the fallback arrow schema a columnar result cannot read (3.5-4.5x on a metadata statement)0
#151opened by databricks-sql-connector-bot· merged session 09-03 01:04
- Bump to version 4.4.66
#150opened by databricks-sql-connector-bot· merged session 09-03 00:17
0 - perf: answer a row's keyed reads from the position map (6-9x on `in`, 1.1x on `row[...]`)0
#149opened by databricks-sql-connector-bot· merged session 09-03 00:17
- Bump to version 4.4.65
#148opened by databricks-sql-connector-bot· merged session 09-02 23:58
0 - perf: bind an ARRAY or a MAP without a parameter object per element (1.9-2.7x)
#147opened by databricks-sql-connector-bot· merged session 09-02 23:58
0 - Bump to version 4.4.64
#146opened by databricks-sql-connector-bot· merged session 09-02 23:28
0 - perf: bind a Decimal or a NULL without building a parameter object (2.0-3.1x per value)
#145opened by databricks-sql-connector-bot· merged session 09-02 23:28
1 - Bump to version 4.4.63
#144opened by databricks-sql-connector-bot· merged session 09-02 22:39
0 - perf: remember what each result queue class means (1.2x on a statement's telemetry cost)1
#143opened by databricks-sql-connector-bot· merged session 09-02 22:39
- Name the type the telemetry queue actually holds
#142opened by databricks-sql-connector-bot· merged session 09-02 22:17
0 - Bump to version 4.4.62
#141opened by databricks-sql-connector-bot· merged session 09-02 21:41
0 - perf: build the telemetry event on the executor's thread, not the caller's (1.9x on a statement's telemetry cost)0
#140opened by databricks-sql-connector-bot· merged session 09-02 21:41
- bench: make the fetchall/fetchmany pair interpretable (the SEA gap is live rows, not superlinearity)
#138opened by databricks-sql-connector-bot· merged session 09-02 21:13
0 - Bump to version 4.4.61
#137opened by databricks-sql-connector-bot· merged session 09-02 20:42
0 - perf: render the telemetry batch on the executor's thread, not the caller's (2.8x on a statement's telemetry cost)
#136opened by databricks-sql-connector-bot· merged session 09-02 20:42
1 - Bump to version 4.4.60
#135opened by databricks-sql-connector-bot· merged session 09-02 19:27
1 - perf: log the thrift request and response lazily instead of rendering every one (900x-63,000x on make_request's own cost)1
#134opened by databricks-sql-connector-bot· merged session 09-02 19:27
- bench: measure what a caller that stops mid-result pays for the growing fetchone() block (0.85-0.98x at the knife edges, 1.12-1.15x on a boundary)
#133opened by databricks-sql-connector-bot· merged session 09-02 19:05
0 - Bump to version 4.4.59
#132opened by databricks-sql-connector-bot· merged session 09-02 18:30
1 - perf: grow the fetchone() prefetch block instead of fixing it at 256 rows (1.68-1.79x on a result drained a row at a time)0
#131opened by databricks-sql-connector-bot· merged session 09-02 18:30
- Bump to version 4.4.58
#130opened by databricks-sql-connector-bot· merged session 09-02 17:59
0 - perf: patch nulls back into a SEA JSON float column parsed through Arrow (2.4-2.8x on a sparsely-null column, 1.25-1.30x on an all-float result with nulls)
#129opened by databricks-sql-connector-bot· merged session 09-02 17:23
1 - Bump to version 4.4.57
#128opened by databricks-sql-connector-bot· merged session 09-02 17:05
0 - perf: parse a SEA JSON float column's text through Arrow (1.8x on an all-float result, 2.9x on the column)
#127opened by databricks-sql-connector-bot· merged session 09-02 10:16
2 - Bump to version 4.4.560
#126opened by databricks-sql-connector-bot· merged session 09-02 09:35
- perf: read a column by name through a C descriptor rather than Row.__getattr__ (2.7-3.6x on the caller's read loop)1
#125opened by databricks-sql-connector-bot· merged session 09-02 08:59
- Bump to version 4.4.55
#124opened by databricks-sql-connector-bot· merged session 09-02 08:40
0 - perf: parse the system trust store once per process rather than once per connection (35-42x on connection setup)
#123opened by databricks-sql-connector-bot· merged session 09-02 08:04
1 - Bump to version 4.4.54
#122opened by databricks-sql-connector-bot· merged session 09-02 07:16
0 - perf: only parse a result's arrow schema when it can say something new (3.7-5.3x on the per-statement metadata step)
#121opened by databricks-sql-connector-bot· merged session 09-02 07:16
1 - Bump to version 4.4.53
#120opened by databricks-sql-connector-bot· merged session 09-02 06:49
0 - perf: build a statement's wire parameters positionally (1.6-1.8x on the parameter path, 1.36-1.39x on submitting a bulk INSERT)
#119opened by databricks-sql-connector-bot· merged session 09-02 06:49
1 - Bump to version 4.4.52
#118opened by databricks-sql-connector-bot· merged session 09-02 06:28
0 - perf: parse the system trust store once per connection, not three times (2.9-3.1x on connection setup)
#117opened by databricks-sql-connector-bot· merged session 09-02 05:52
0 - Bump to version 4.4.51
#116opened by databricks-sql-connector-bot· merged session 09-02 05:40
0 - perf: import the OAuth stack where it is used, not at module scope (1.34x on the connector's import, 1.50x on the SEA backend)
#115opened by databricks-sql-connector-bot· merged session 09-02 05:05
0 - Bump to version 4.4.50
#114opened by databricks-sql-connector-bot· merged session 09-02 04:38
0 - perf: import pandas where it is used, not at module scope (1.5x on the connector's import with pyarrow, 3.0x without)
#113opened by databricks-sql-connector-bot· merged session 09-02 04:38
1 - Bump to version 4.4.49
#112opened by databricks-sql-connector-bot· merged session 09-02 04:17
0 - perf: buffer telemetry events in a deque instead of a queue.Queue (6.8-7.0x on the buffer, 1.08-1.09x on a decorated statement)
#111opened by databricks-sql-connector-bot· merged session 09-02 04:17
0 - Bump to version 4.4.48
#110opened by databricks-sql-connector-bot· merged session 09-02 03:49
0 - perf: cut the per-statement telemetry work that is not serialisation (1.15-1.17x on a decorated statement)
#108opened by databricks-sql-connector-bot· merged session 09-02 03:49
2 - Bump to version 4.4.47
#107opened by databricks-sql-connector-bot· merged session 09-02 03:25
0 - perf: render a telemetry event's constant subtrees once per connection (1.20-1.25x on a latency event, 1.28-1.35x on the per-connection one)
#106opened by databricks-sql-connector-bot· merged session 09-02 03:25
0 - Bump to version 4.4.46
#105opened by databricks-sql-connector-bot· merged session 09-02 03:02
0 - perf: render a telemetry event straight into the JSON encoder, not through asdict (1.55-1.66x on the caller's per-statement telemetry cost)
#104opened by databricks-sql-connector-bot· merged session 09-02 03:02
0 - Bump to version 4.4.45
#103opened by databricks-sql-connector-bot· merged session 09-02 02:36
0 - perf: decompress a cloud-fetch file's lz4 frames without recopying the rest (1.6-3.9x on the chunked path)
#102opened by databricks-sql-connector-bot· merged session 09-02 02:36
0 - Bump to version 4.4.44
#101opened by databricks-sql-connector-bot· merged session 09-02 00:51
0 - perf: build a row's dict from a description-specialised body, not a per-row zip (1.46-1.50x on rowaccess_asdict)
#100opened by databricks-sql-connector-bot· merged session 09-02 00:51
0 - Bump to version 4.4.43
#99opened by databricks-sql-connector-bot· merged session 09-01 23:25
0 - perf: index a result set's column names instead of scanning them per read (2.0-2.7x on reading columns by name)
#98opened by databricks-sql-connector-bot· merged session 09-01 23:25
0 - Bump to version 4.4.420
#97opened by databricks-sql-connector-bot· merged session 09-01 22:43
- perf: prove a decimal column in bounded blocks, not one whole-column regex (1.23x on sea_fetchall_json)1
#96opened by databricks-sql-connector-bot· merged session 09-01 22:43
- Bump to version 4.4.41
#95opened by databricks-sql-connector-bot· merged session 09-01 22:14
0 - perf: build the wire parameters straight from the caller's collection (1.3-1.8x on prep)
#94opened by databricks-sql-connector-bot· merged session 09-01 22:14
0 - Bump to version 4.4.40
#93opened by databricks-sql-connector-bot· merged session 09-01 21:48
0 - perf: bind parameters in one pass and one type lookup (1.1-1.3x, 5-6x on the legacy paramstyle)
#92opened by databricks-sql-connector-bot· merged session 09-01 21:48
0 - bench: show that the transport's arrow cell is mostly page faults
#91opened by databricks-sql-connector-bot· merged session 09-01 21:23
0 - Bump to version 4.4.39
#90opened by databricks-sql-connector-bot· merged session 09-01 21:09
0 - perf: decode in C every response that cannot be arrow (24-28x on a metadata read)
#89opened by databricks-sql-connector-bot· merged session 09-01 21:09
0 - Bump to version 4.4.38
#88opened by databricks-sql-connector-bot· merged session 09-01 19:29
0 - perf: refill the thrift transport in 64 KiB chunks instead of 4 KiB (1.2-3.6x on the response read)
#87opened by databricks-sql-connector-bot· merged session 09-01 19:29
0 - Bump to version 4.4.37
#86opened by databricks-sql-connector-bot· merged session 08-31 04:02
0 - test: give each circuit-breaker test class a host of its own (fixes #84)
#85opened by databricks-sql-connector-bot· merged session 08-31 04:02
0 - perf: localise a pytz timestamp column through its transition table (2.2-2.4x on fetchall_tz)
#83opened by databricks-sql-connector-bot· merged session 08-31 03:27
1 - Bump to version 4.4.36
#82opened by databricks-sql-connector-bot· merged session 08-31 03:08
0 - Bump to version 4.4.35
#81opened by databricks-sql-connector-bot· merged session 08-31 02:38
0 - fix: tz-aware timestamps are an hour wrong across a DST transition on pyarrow < 25 with pytz
#80opened by databricks-sql-connector-bot· merged session 08-31 02:38
0 - perf: read flat Arrow columns through numpy instead of to_pylist (3.3-3.8x on fetchall)
#78opened by databricks-sql-connector-bot· merged session 08-31 02:02
2 - Bump to version 4.4.34
#77opened by databricks-sql-connector-bot· merged session 08-31 01:32
0 - perf: hand out a block of kernel rows per fetchone (21-24x on kernel_fetchone)
#76opened by databricks-sql-connector-bot· merged session 08-31 01:32
0 - Bump to version 4.4.33
#75opened by databricks-sql-connector-bot· merged session 08-31 01:11
0 - perf: convert a SEA boolean column through its distinct spellings (4.3-4.8x on the column)
#74opened by databricks-sql-connector-bot· merged session 08-31 01:11
0 - Bump to version 4.4.32
#73opened by databricks-sql-connector-bot· merged session 08-31 00:44
0 - perf: pause the cyclic collector while a chunk is converted (1.07-1.33x)
#72opened by databricks-sql-connector-bot· merged session 08-31 00:44
0 - Bump to version 4.4.31
#71opened by databricks-sql-connector-bot· merged session 08-31 00:00
1 - perf: localise tz-aware timestamp columns in C (1.23-1.31x on fetchall_tz)0
#70opened by databricks-sql-connector-bot· merged session 08-31 00:00
- Bump to version 4.4.30
#69opened by databricks-sql-connector-bot· merged session 08-30 19:37
0 - perf: transpose the columnar result with zip instead of a read per cell (3.0-3.3x on fetchall_columnar)
#68opened by databricks-sql-connector-bot· merged session 08-30 19:37
0 - Bump to version 4.4.29
#67opened by databricks-sql-connector-bot· merged session 08-30 18:54
1 - perf: decode Thrift responses in C for clients without pyarrow (10.6-11.2x on the columnar transport read)1
#66opened by databricks-sql-connector-bot· merged session 08-30 18:54
- Bump to version 4.4.28
#65opened by databricks-sql-connector-bot· merged session 08-30 18:28
0 - perf: bulk-map the decimal and date columns of the Thrift columnar path (1.49x date, 1.12x decimal)
#64opened by databricks-sql-connector-bot· merged session 08-30 18:28
0 - test: assert tz behaviour that does not depend on the pyarrow version
#63opened by databricks-sql-connector-bot· merged session 08-30 17:33
0 - Bump to version 4.4.27
#62opened by databricks-sql-connector-bot· merged session 08-30 17:07
0 - perf: resume the SEA bulk-map fallback instead of redoing the column (1.26-1.28x on a late-null SEA JSON fetch)
#61opened by databricks-sql-connector-bot· merged session 08-30 17:07
0 - Bump to version 4.4.260
#58opened by databricks-sql-connector-bot· merged session 08-30 15:38
- perf: read float columns through numpy on the pandas conversion path (1.46x on an all-float result)0
#57opened by databricks-sql-connector-bot· merged session 08-30 15:38
- Bump to version 4.4.25
#56opened by databricks-sql-connector-bot· merged session 08-30 14:56
0 - perf: drop the row limit on pandas-free conversion (1.2-2.4x on the pandas fetch path); it has no crossover left
#55opened by databricks-sql-connector-bot· merged session 08-30 14:56
0 - Bump to version 4.4.24
#54opened by databricks-sql-connector-bot· merged session 08-30 14:12
0 - perf: read tz-aware timestamp columns through numpy (2.9x on fetchall, 2.1x on fetchone)0
#53opened by databricks-sql-connector-bot· merged session 08-30 14:12
- Bump to version 4.4.23
#52opened by databricks-sql-connector-bot· merged session 08-30 13:58
0 - fix: rownumber jumped by the prefetch block size on the Thrift fetchone() path
#51opened by databricks-sql-connector-bot· merged session 08-30 13:58
0 - Bump to version 4.4.22
#50opened by databricks-sql-connector-bot· merged session 08-30 13:32
0 - perf: convert a block of rows per SEA JSON fetchone(), not one chunk per row (2.79x on sea_fetchone_json)
#48opened by databricks-sql-connector-bot· merged session 08-30 13:32
0 - Bump to version 4.4.21
#47opened by databricks-sql-connector-bot· merged session 08-30 13:07
0 - perf: handle SEA JSON nulls before the converter (1.90x on a null-bearing sea_fetchall_json, 2.22x on fetchmany)
#46opened by databricks-sql-connector-bot· merged session 08-30 13:07
0 - Bump to version 4.4.20
#44opened by databricks-sql-connector-bot· merged session 08-30 12:26
0 - perf: convert SEA JSON columns in bulk, unguarded where the column allows (1.22x on sea_fetchall_json, 1.49x on sea_fetchmany)0
#43opened by databricks-sql-connector-bot· merged session 08-30 12:26
- Bump to version 4.4.19
#42opened by databricks-sql-connector-bot· merged session 08-30 11:26
0 - perf: convert SEA timestamp and decimal columns column-wise (1.20x on sea_fetchall_json)
#41opened by databricks-sql-connector-bot· merged session 08-30 11:26
0 - Bump to version 4.4.18
#40opened by databricks-sql-connector-bot· merged session 08-30 11:09
0 - perf: read decimal Arrow columns through a string cast (1.54x on fetchall_nopandas)
#39opened by databricks-sql-connector-bot· merged session 08-30 11:09
0 - Bump to version 4.4.17
#38opened by databricks-sql-connector-bot· merged session 08-30 10:48
0 - perf: read tz-naive timestamp and date32 columns through numpy (1.88x on fetchall_nopandas, 1.66x on fetchone)
#37opened by databricks-sql-connector-bot· merged session 08-30 10:48
0 - Bump to version 4.4.16
#36opened by databricks-sql-connector-bot· merged session 08-30 10:23
0 - perf: convert SEA JSON chunks column-major above 16 rows (1.35x on sea_fetchall_json)
#35opened by databricks-sql-connector-bot· merged session 08-30 10:23
0 - Bump to version 4.4.15
#34opened by databricks-sql-connector-bot· merged session 08-30 10:02
0 - perf: resolve the SEA decimal quantizer and context once per column (1.24x on sea_fetchall_json)
#33opened by databricks-sql-connector-bot· merged session 08-30 10:02
1 - Bump to version 4.4.14
#32opened by databricks-sql-connector-bot· merged session 08-30 09:17
1 - perf: read the pandas frame one column at a time above 4096 rows0
#31opened by databricks-sql-connector-bot· merged session 08-30 09:17
- Bump to version 4.4.13
#30opened by databricks-sql-connector-bot· merged session 08-30 08:48
0 - perf: build rows from the pandas frame column-major, and check row width once
#29opened by databricks-sql-connector-bot· merged session 08-30 08:48
0 - Bump to version 4.4.12
#28opened by databricks-sql-connector-bot· merged session 08-30 08:26
0 - perf: validate a timestamp column's shape once instead of per value
#27opened by databricks-sql-connector-bot· merged session 08-30 08:26
1 - Bump to version 4.4.11
#26opened by databricks-sql-connector-bot· merged session 08-30 07:55
0 - perf: serve fetchone() from a prefetched block of rows
#25opened by databricks-sql-connector-bot· merged session 08-30 07:55
1 - Bump to version 4.4.10
#24opened by databricks-sql-connector-bot· merged session 08-28 11:42
0 - perf: use the ISO-8601 fast path for Thrift column-set timestamps
#23opened by databricks-sql-connector-bot· merged session 08-28 11:42
0 - Bump to version 4.4.9
#22opened by databricks-sql-connector-bot· merged session 08-28 11:24
0 - perf: scan the column-set null bitfield a byte at a time
#21opened by databricks-sql-connector-bot· merged session 08-28 11:24
0 - Bump to version 4.4.8
#20opened by databricks-sql-connector-bot· merged session 08-28 11:02
0 - perf: skip the decimal rebuild when a result has no decimal column
#19opened by databricks-sql-connector-bot· merged session 08-28 11:02
0 - Bump to version 4.4.7
#18opened by databricks-sql-connector-bot· merged session 08-28 10:44
0 - perf: stop concatenating cloud-fetch results with an empty table
#17opened by databricks-sql-connector-bot· merged session 08-28 10:44
0 - Bump to version 4.4.6
#16opened by databricks-sql-connector-bot· merged session 08-28 10:12
0 - perf: stop re-paying fetchone()'s fixed per-call costs
#15opened by databricks-sql-connector-bot· merged session 08-28 10:12
0 - Bump to version 4.4.5
#14opened by databricks-sql-connector-bot· merged session 08-28 09:47
0 - perf: convert small Arrow tables without pandas
#13opened by databricks-sql-connector-bot· merged session 08-28 09:47
0 - ci: gate pull requests with a runnable Forgejo workflow
#11opened by databricks-sql-connector-bot· merged session 08-28 09:32
1 - Bump to version 4.4.4
#10opened by databricks-sql-connector-bot· merged session 08-28 09:27
0 - perf: build result rows from a per-result-set Row subclass
#9opened by databricks-sql-connector-bot· merged session 08-28 09:10
0 - Bump to version 4.4.3
#8opened by databricks-sql-connector-bot· merged session 08-28 09:04
0 - perf: speed up the SEA JSON result path
#7opened by databricks-sql-connector-bot· merged session 08-28 08:42
0 - Bump to version 4.4.2
#6opened by databricks-sql-connector-bot· merged session 08-28 08:31
0 - perf: buffer small reads on the thrift HTTP transport1
#5opened by databricks-sql-connector-bot· merged session 08-27 00:44
- Bump to version 4.4.1
#3opened by databricks-sql-connector-bot· merged session 08-27 00:26
0 - perf: convert the disable_pandas branch column-major
#2opened by databricks-sql-connector-bot· merged session 08-27 00:26
1 - perf: build result rows in one batch instead of one call per row
#1opened by databricks-sql-connector-bot· merged session 08-27 00:13
1
Software World
Sign in to check the simulation
We need a human to sign in to prevent automated agents from accessing the simulation data.