Skip to content

v0.88.0-beta.2

Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 21 Nov 00:30
· 5 commits to main since this release
Immutable release. Only release title and notes can be modified.
v0.88.0-beta.2
8ac7164

Added

  • Share a query, a saved query or a result as a link anyone can open in a browser. (#4253)
  • Share Result masks columns that look sensitive before upload, and lets you keep, mask or remove each column. (#4253)
  • Open a shared link in TablePro to get its SQL in a new query tab, without running it. (#4253)
  • Select several rows on iPhone and iPad to export, copy or delete them. (#4302)
  • Import CSV, JSON and XLSX files into a table, and run SQL files, on iPhone and iPad. (#4302)
  • Export tables, query results, selected rows or a whole database to a file on iPhone and iPad. (#4302)
  • Drag an editor tab onto another window of the same connection to move it there. (#4323)
  • Name a query tab by double-clicking it or choosing Rename Tab, and the name stays after a restart. (#4343)
  • Open a cell's value editor from its context menu, the Query menu, or Option-Return. (#4374)
  • Double-click a PHP serialized value in a read-only result to open it in the viewer. (#4377)
  • Wrap Text in the data grid shows long values on several lines, up to 10 lines a row. (#4378)

Changed

  • Typing in the query editor no longer redraws the tab strip and the rest of the window on every keystroke. (#4286)
  • Restore Dump's database picker shows favorite databases and their menu, like the toolbar switcher. (#4303)
  • The editor tab bar now scrolls, animates and drags like the macOS tab bar. (#4323)
  • Fetch All on PostgreSQL uses less memory, and binary columns load much faster. (#4328)
  • TLS connections to PostgreSQL 17 and later skip a round trip once the server is known. (#4328)
  • PostgreSQL connections open faster and a table's columns load in one round trip. (#4328)
  • Row totals stay in place while you page through a table, until you refresh or change the filters. (#4330)
  • Spanner, Cloudflare D1 and SurrealDB tables are counted only when you choose Count Exactly. (#4330)
  • SSH tunnels transfer large results many times faster over high-latency connections. (#4331)
  • libSQL queries stop at the row limit instead of downloading the whole result first. (#4331)
  • Snowflake table columns and details load without a running warehouse. (#4331)
  • Oracle results that stream in fetch 1,000 rows per round trip instead of 50. (#4331)
  • Large ClickHouse results decode about 2.7 times faster. (#4331)
  • Elasticsearch deep pages no longer download the documents they skip. (#4332)
  • Memcached key browsing reads a page of values in one request. (#4332)
  • Redis key browsing loads a page in fewer round trips on databases with 10,000 keys or more. (#4332)
  • MongoDB collection pages load without an extra round trip to the server. (#4332)
  • CSV and JSON exports of large tables finish several times faster. (#4333)
  • Refreshing a database with thousands of tables now updates only the sidebar rows that changed. (#4334)
  • VoiceOver reads the data grid columns on screen, and scrolling sideways brings in the next ones. (#4335)
  • Opening a table redraws the data grid once instead of several times. (#4335)
  • Selecting hundreds of thousands of rows in the data grid stays responsive. (#4335)
  • Wide tables and long values draw and scroll faster in the data grid. (#4335)
  • Large results use less memory, and the first edit in one no longer pauses to copy every row. (#4340)
  • Typing on a long line with line wrapping off no longer lays out every visible line again. (#4340)
  • Typing in a very large SQL file no longer pauses while syntax highlighting catches up. (#4340)
  • ROW_COUNT() and SHOW WARNINGS on MySQL and MariaDB answer for your last statement, not for a database switch. (#4341)
  • A MySQL or MariaDB query with a row limit sends the limit together with the query. (#4341)
  • MySQL and MariaDB no longer switch database before each statement, so paging and running queries take half the time. (#4341)
  • Opening a MySQL or MariaDB table reads its columns in one round trip. (#4341)
  • PostgreSQL and SQL Server results take less CPU to read. (#4342)
  • Table tabs open sorted by their primary key unless you chose another default row sort. (#4344)
  • The first page of a Kafka topic loads with one request per broker instead of one per partition. (#4349)
  • Cassandra table pages after the first load without preparing the query again. (#4349)
  • The Data Dictionary export reads a PostgreSQL schema's columns, keys and indexes in one pass, not table by table. (#4350)
  • BigQuery statements that finish within two seconds return in one request, without the half-second wait each one had. (#4352)
  • DuckDB connections now refuse SET threads below the number of processors. (#4354)
  • A PostgreSQL SQL export reads each schema's indexes, comments, sequences and enum types in one pass, not table by table. (#4355)
  • A PostgreSQL SQL import with Wrap in transaction on sends its writes in batches, not one statement at a time. (#4356)
  • A PostgreSQL SQL import stops at a statement read before an earlier write changed standard_conforming_strings. (#4356)
  • Tables open a round trip sooner from the columns read with the schema, and a bar offers Reload if they changed since. (#4357)

Removed

  • Removed the setting that wrapped editor tabs onto more rows. (#4323)

Fixed

  • Switching tabs cleared the undo history of grid edits still waiting to be saved. (#4283)
  • One connection TablePro could not read emptied the whole connection list and blocked saving. (#4283)
  • After a reconnect lost a transaction, a grid save used up the warning meant for your next statement. (#4283)
  • Choosing Save when closing a tab could save another tab instead if the selection changed while the prompt was up. (#4286)
  • Query History saved the statements of a script without their schema, so reopening one ran in the default schema. (#4286)
  • Query Insights grouped statements without the engine's quoting and comment rules, such as PostgreSQL's 'C:\'. (#4289)
  • Importing a SQL file joined the words around a dropped comment, so a/* x */FROM t ran as aFROM t. (#4289)
  • Importing a SQL file split a PostgreSQL E'...' string continued on the next line, or a SQLite $name(...) parameter. (#4289)
  • The editor flagged a bracket inside a $$ body, a [name], a MySQL # comment or PostgreSQL's 'C:\' as unmatched. (#4289)
  • On iOS, a ROLLBACK after a MySQL # comment or a nested PostgreSQL comment left the transaction shown as open. (#4289)
  • Renaming a SQL Server column failed, or renamed the wrong one, when a schema or table name held a dot or a bracket. (#4294)
  • XLSX export dropped leading and trailing spaces in text cells. (#4302)
  • SQLite SQL exports dropped NUL characters inside text. (#4302)
  • SQL exports wrote wrong trigger or routine drops for Cassandra, Teradata, SQL Server aggregates and DuckDB macros. (#4302)
  • Oracle triggers were left out of SQL exports, and Oracle drops failed to restore before 23ai. (#4302)
  • An XLSX file with one huge shared string could use up memory while it was previewed. (#4302)
  • A SQL export of query results failed with Read every table at one snapshot turned on. (#4302)
  • CSV, JSON and XLSX import was offered for MongoDB and then failed. (#4302)
  • Inserting a row of only defaults on Oracle did nothing on iPhone and iPad. (#4302)
  • Copy as SQL INSERT on iPhone and iPad quoted MySQL BIT values and listed generated columns. (#4302)
  • A small SQL file whose last characters were Latin-1 was read as UTF-8. (#4302)
  • Stopping an XLSX or Markdown export after its last row still saved the file. (#4302)
  • SQL exports with Drop wrote DROP ROUTINE, which no database accepts, and dropped PostgreSQL procedures as functions. (#4302)
  • SQL exports from two windows at once could leave out sequences and mix up their warnings. (#4302)
  • Markdown export cut emoji and dropped accents in aligned columns. (#4302)
  • PostgreSQL booleans showed as t and f on iPhone and iPad. (#4302)
  • A Claude, Gemini, OpenAI or xAI reply that ended with nothing in it showed an empty message instead of an error. (#4317)
  • PostgreSQL pages of a large table with no sort order no longer skip rows. (#4328)
  • Paging through a table without a primary key no longer rereads its structure on every page. (#4330)
  • Views and tables with no row estimate no longer run a full row count when you open them. (#4330)
  • Clicking another table no longer waits for the previous table's structure to finish loading. (#4330)
  • Snowflake clustering keys no longer come from a table whose name differs only by case. (#4331)
  • Parquet exports of large tables no longer take most of the memory on your Mac. (#4333)
  • On iPhone and iPad, one connection, group or tag the app could not read emptied the library and blocked changes. (#4339)
  • After a dropped MySQL connection, a read could rerun on the wrong database when USE was typed without a space. (#4341)
  • Opening a new tab crashed when another tab was titled Query 9223372036854775807. (#4343)
  • A saved or shared tab titled Query 2 was renumbered when moved to another connection or restored. (#4343)
  • Renaming a database or schema replaced the SQL in a query tab whose last run read a table there. (#4343)
  • After renaming a table or schema, the tab and the window title showed different names. (#4343)
  • Tabs such as ER Diagram kept their old language after you changed the app's language. (#4343)
  • A reopened tab could share its Query number with another open tab. (#4343)
  • A tab saved with Save As showed its file name without the extension, unlike the same file opened later. (#4343)
  • After a COMMIT, autocomplete offered no columns before FROM until it read them again. (#4344)
  • A saved connection the app could not read was replaced by its iCloud copy, or stayed after being deleted elsewhere. (#4345)
  • In Vim mode, marks stayed where they were when lines were typed, pasted or undone above them. (#4347)
  • In Vim mode, cw, cj, dF and db at the start of a line changed more text than Vim does. (#4347)
  • In Vim mode, a and p on an empty line put the text on the next line. (#4347)
  • In Vim mode, "a2p pasted register a once and then the last yank, and 2P mixed up the copies. (#4347)
  • In Vim mode, d%, c%, dn, d}, d'a and other motions after an operator only moved the cursor. (#4347)
  • In Vim mode, Escape or leaving the command line kept a pending d or c, so the next key could delete text. (#4347)
  • In Vim mode, ip dropped the last character of the last paragraph. (#4347)
  • In Vim mode, Enter, + and - did nothing in Normal mode. (#4347)
  • In Vim mode, w stopped on a symbol followed by a variation selector, such as ❤️. (#4347)
  • In Vim mode, the next motion after viw shrank the selection instead of extending it. (#4347)
  • A PostgreSQL dump of a table using one enum type in two columns listed each label twice, so its CREATE TYPE failed. (#4355)
  • A SQL export or Data Dictionary dropped or mixed up tables and schemas whose names differ only in Unicode form. (#4359)
  • A DuckDB SET or table function that calls nextval() no longer breaks the connection until it is reopened. (#4364)
  • A SQL export dropped values or mixed up keys when two tables, columns or keys had names differing only in Unicode form. (#4365)
  • The ER diagram, sidebar, autocomplete and compare merged tables or schemas named alike in two Unicode forms. (#4366)
  • Pressing Return or typing on a checkbox cell opened a text field over the checkbox. (#4374)
  • A Windows path or other text that only starts like a PHP token was shown as an invalid PHP value. (#4377)

Security

  • Opening or browsing a DuckDB file with deeply nested constraints, defaults or views could crash the app. (#4293)
  • Attaching a DuckDB file with deeply nested constraints, defaults or views could crash the app. (#4354)