;update changelogs

CHANGES.md

@@ -1,15 +1,21 @@
 General/project-related changes in the hledger project. 
 For package-specific changes, see the package changelogs.
 
-# 093c11ec
+# f39938d7
 
+- examples: stripe csv
 
+- The functional tests in tests/ have been moved into the respective
+  packages, eg hledger/test/ and hledger-ui/test/.
 
-- benchmarks: run just the slowest commands; add some large tabular reports
+- Shake cabalfiles: now gives an error when it fails
 
-- tools: use pretty-simple for ghci output
+- make bench: add some large tabular reports; 
+  run just the slowest commands by default;
+  run after make (func)test
 
-- tools: add hie.yaml, hls now works with hledger
+- a hie.yaml file has been added, so hledger source loads
+  easily in IDEs supporting haskell-language-server
 
 
 # 1.19.1 2020-09-07

hledger-lib/CHANGES.md

@@ -1,7 +1,24 @@
 Internal/api/developer-ish changes in the hledger-lib (and hledger) packages.
 For user-visible changes, see the hledger package changelog.
 
-# 093c11ec
+# 3662977c
+
+- Reverted a stripAnsi change in 1.19.1 that caused a 3x slowdown of amount rendering 
+  in terminal reports. (#1350)
+
+- Amount and table rendering has been improved, so that stripAnsi is no longer needed.
+  This speeds up amount rendering in the terminal, speeding up some reports by 10% or more since 1.19.
+  (Stephen Morgan)
+
+- global commodity display styles can now be set in InputOpts or Journal,
+  overriding all others (declared or inferred). This is used by the import
+  command and probably command-line options in future.
+
+- Journal keeps a new piece of parsing state, a decimal mark character,
+  which can optionally be set to force the number format expected by all
+  amount parsers.
+
+- Remove Empty Query constructor, which does nothing and has done so for a very long time. (Stephen Morgan)
 
 - In ReportOpts, store query terms term-by-term in a list in querystring_. (Stephen Morgan)
   This helps deal with tricky quoting issues, as we no longer have to make
@@ -25,73 +42,81 @@ For user-visible changes, see the hledger package changelog.
 
 - Export some MultiBalanceReport helper functions. (Stephen Morgan)
 
-- Introduce ReportSpec, which holds ReportOpts, the day of the report, and the parsed Query. (Stephen Morgan)
-
-- Remove old impure ReportOpts date functions. (Stephen Morgan)
-
 - Make Default instances clearer, remove Default instance for Bool. (Stephen Morgan)
 
-- Store the original query string in ReportOpts, provide a function for regenerating ReportOpts. (Stephen Morgan)
+- Many ReportOpts-related changes, such as the addition of ReportSpec, aimed
+  at preventing runtime errors (from parsing: regexps, dates, format strings;
+  from not having today's date set; etc.)
+  ReportSpec holds a ReportOpts, the day of the report, and the Query generated from these.
 
-- Ensure ReportOpts always has today_ set. (Stephen Morgan)
+- StringFormat now takes an optional overline width, which is
-
+  currently only used by defaultBalanceLineFormat. (Stephen Morgan)
-- Make sure reportspan doesn't interfere with correctly determining valuation date. (Stephen Morgan)
-
-- Store parsed Query in ReportOpts, rather than an unparsed String. (Stephen Morgan)
-
-- Store StringFormat in ReportOpts, rather than unparsed String. (Stephen Morgan)
-  StringFormat now also takes an optional overline width, which is
-  currently only used by defaultBalanceLineFormat.
-
-- Remove checkReportOpts and checkRawOpts. (Stephen Morgan)
-  checkRawOpts has been a no-op for at least four years, and
-  checkReportOpts only makes sure that depth_ is positive, which is taken
-  care of by the maybeposintopt parser.
-
-- For ymd date parsing, don't consume invalid date components. (Stephen Morgan)
 
 - quoteIfNeeded should not escape the backslashes in unicode code points. (Stephen Morgan)
 
-- fix error when P directive has a zero price  (#1373)
-
 - Export OrdPlus and constructors. (Stephen Morgan)
 
-- fix a slowdown with report rendering in 1.19.1 (#1350)
+- Debug output now uses pretty-simple instead pretty-show.
-  stripAnsi is called many times during rendering (by strWidth), so
+  This hopefully gives overall nicer debug output (eg in colour), 
-  should be fast. It was originally a regex replacement, and more
+  including for values which don't have Read-able Show output.
-  recently a custom parser. The parser was slower, particularly the one
+  This means that we can start removing custom Show instances 
-  in 1.19.1. See #1350, and this rough test:
+  that were a workaround for pretty-show. Eg account names
+  in debug output no longer show their colons as underscores.
 
-  time118ish = timeIt $ print $ length $ concat $ map (fromRight undefined . regexReplace (toRegex' "\ESC\\[([0-9]+;)*([0-9]+)?[ABCDHJKfmsu]") "") testdata
+  Here's some old pretty-show output:
-  time119    = timeparser (many (takeWhile1P Nothing (/='\ESC') <|> "" <$ ansi))
-  time1191   = timeparser (many ("" <$ try ansi <|> pure <$> anySingle))
-  timeparser p = timeIt $ print $ length $ concat $ map (concat . fromJust . parseMaybe p) testdata
-  testdata = concat $ replicate 10000
-      [ "2008-01-01 income               assets:bank:checking            $1            $1"
-      , "2008-06-01 gift                 assets:bank:checking            $1            $2"
-      , "2008-06-02 save                 assets:bank:saving              $1            $3"
-      , "                                assets:bank:checking  ..m$-1\ESC[m\ESC[m            $2"
-      , "2008-06-03 eat & shop           assets:cash           ..m$-2\ESC[m\ESC[m             0"
-      , "2008-12-31 pay off              assets:bank:checking  ..m$-1\ESC[m\ESC[m  ..m$-1\ESC[m\ESC[m"
-      ]
 
-  ghci> time118ish
+   CsvRules
-  4560000
+     { rdirectives = [ ( "skip" , "1" ) ]
-  CPU time:   0.17s
+     , rcsvfieldindexes = [ ( "date" , 1 ) , ( "amount" , 2 ) ]
-  ghci> time119
+     , rassignments = [ ( "amount" , "%2" ) , ( "date" , "%1" ) ]
-  4560000
+     , rconditionalblocks = []
-  CPU time:   0.91s
+     }
-  ghci> time1191
-  4560000
-  CPU time:   2.76s
 
-  Possibly a more careful parser could beat regexReplace. Note the
+  And the new pretty-simple output:
-  latter does memoisation, which could be faster and/or could also use
-  more resident memory in some situations.
 
-  Ideally we would calculate all widths before adding ANSI colour codes,
+   CsvRules
-  so we wouldn't have to wastefully strip them.
+     { rdirectives=
+       [ ( "skip", "1" ) ]
+     , rcsvfieldindexes=
+       [ ( "date", 1 ), ( "amount", 2 ) ]
+     , rassignments=
+       [ ( "amount", "%2" ), ( "date", "%1" ) ]
+     , rconditionalblocks= []
+     }
 
+  We require pretty-simple 4.0.0.0 to get this compact output.
+  It's a little less compact than pretty-show, but not too bad.
+  Non-compact pretty-simple output would be:
+
+   CsvRules
+       { rdirectives=
+           [
+               ( "skip"
+               , "1B"
+               )
+           ]
+       , rcsvfieldindexes=
+           [
+               ( "date"
+               , 1
+               )
+           ,
+               ( "amount"
+               , 2
+               )
+           ]
+       , rassignments=
+           [
+               ( "amount"
+               , "%2"
+               )
+           ,
+               ( "date"
+               , "%1"
+               )
+           ]
+       , rconditionalblocks=[]
+       }
 
 
 # 1.19.1 2020-09-07

hledger-ui/CHANGES.md

@@ -1,13 +1,21 @@
 User-visible changes in hledger-ui.
 See also the hledger changelog.
 
-# 093c11ec
+# 3662977c
+
+- When entering a query with `/`, malformed queries/regular expressions
+  no longer cause the program to exit. (Stephen Morgan)
 
 - Eliding of multicommodity amounts now makes better use of available space. (Stephen Morgan)
 
-- E ignores file extension, should help positioning on windows
+- `E` now parses the `HLEDGER_UI_EDITOR` or `EDITOR` environment variable
+  correctly on Windows (ignoring the file extension), so if you have that set
+  it should be better at opening your editor at the correct line.
 
-- E supports positioning when EDITOR is code (VS Code) (#1359)
+- `E` now supports positioning when `HLEDGER_UI_EDITOR` or `EDITOR` 
+  is VS Code ("`code`") (#1359)
+
+- hledger-ui now has a (human-powered) test suite.
 
 
 # 1.19.1 2020-09-07

hledger-web/CHANGES.md

@@ -1,18 +1,19 @@
 User-visible changes in hledger-web.
 See also the hledger changelog.
 
-# 093c11ec
+# 3662977c
 
+- Re-enable hledger-web's test suite, and include it in the main executable.
+  hledger-web --test [-- HSPECARGS] runs it.
 
+- Fix --forecast, broken in hledger-web since 1.18 (#1390)
 
-- Fix hledger-web description (TANIGUCHI Kohei)
+- Fix unescaped slashes in hledger-web description on hackage  (TANIGUCHI Kohei)
-  Slashes need to be escaped or they introduce unexpected italic style
-  due to Haddock markup.
 
-- The hledger version string, as JSON, is provided at /version (#1152)
+- The hledger-web version string is now provided at /version, as JSON (#1152)
 
-- Write the session file (hledger-web_client_session_key.aes) in 
+- The session file (hledger-web_client_session_key.aes) is now written in 
-  $XDG_DATA_DIR rather than in the current directory.
+  $XDG_DATA_DIR rather than the current directory.
   Eg on non-Windows systems this is ~/.cache/ by default (cf
   https://hackage.haskell.org/package/directory/docs/System-Directory.html#t:XdgDirectory).
   (#1344) (Félix Sipma)

hledger/CHANGES.md

@@ -1,25 +1,69 @@
 User-visible changes in the hledger command line tool and library.
 
 
-# 093c11ec
+# 3662977c
 
-- Query terms containing quotes (eg to match account names containing quotes)
+## general
-  now work properly (#1368, Stephen Morgan)
 
-- Console rendering is more efficient, speeding up some reports by
+- Reverted a stripAnsi change in 1.19.1 that caused a 3x slowdown of amount rendering
-  10% or more (Stephen Morgan)
+  in terminal reports. (#1350)
+
+- Amount and table rendering has been improved, so that stripAnsi is no longer needed.
+  This speeds up amount rendering in the terminal, speeding up some reports by 10% or more since 1.19.
+  (Stephen Morgan)
 
 - Amount eliding no longer displays corrupted ANSI codes (#1352, Stephen Morgan)
 
 - Eliding of multicommodity amounts now makes better use of available space,
-  avoiding unnecessary eliding. (Stephen Morgan)
+  avoiding unnecessary eliding (showing as many amounts as possible within
+  32 characters). (Stephen Morgan)
 
-- --no-elide's help now mentions that it also disables eliding of
+- Command line help for --no-elide now mentions that it also disables eliding of
   multicommodity amounts.
 
-- bal: --budget reports no longer insert an extra space inside the brackets (Stephen Morgan)
+- Query terms containing quotes (eg to match account names containing quotes)
+  now work properly. (#1368, Stephen Morgan)
 
-- journal: forecasted transactions are now affected by commodity styles (#1371)
+- cli, journal: Date range parsing is more robust, fixing failing/incorrect cases such as: (Stephen Morgan)
+
+  - a hyphenated range with just years (`2017-2018`)
+  - a hyphenated date with no day in a hyphenated range (`2017-07-2018`)
+  - a dotted date with no day in a dotted range (`2017.07..2018.02`)
+ 
+- Debug output is prettier (eg, in colour), using pretty-simple instead of pretty-show.
+
+## commands
+
+- add: number style (eg thousands separators) no longer disturbs the value
+  that is offered as default. (#1378)
+
+- bal: --invert now affects -S/--sort-amount, reversing the order. (#1283, #1379) (Stephen Morgan)
+
+- bal: --budget reports no longer insert an extra space inside the brackets. (Stephen Morgan)
+
+- bal, is, bs --change: 
+  Valued multiperiod balance change reports now show changes of value, 
+  rather than the value of changes. (#1353, Stephen Morgan)
+
+- import: The journal's commodity styles (declared or inferred) are now applied
+  to imported amounts, overriding their original number format.
+
+## journal format
+
+- The journal's commodity styles are now applied to forecasted transactions. (#1371)
+
+- journal, csv: commodity style is now inferred from the first amount, as documented,
+  not the last. This was "working wrongly" since hledger 1.12..
+
+- A zero market price no longer causes "Ratio has zero denominator" error
+  in valued reports. (#1373)
+
+## csv format
+
+- The new `decimal-mark` rule allows reliable number parsing
+  when CSV numbers contain digit group marks (eg thousands separators).
+
+- The CSV reader's verbose "assignment" debug output is now at level 9.
 
 
 # 1.19.1 2020-09-07