David,
This fixes the bug you found today in amend-record. It seems that the
type signature we wanted for with_selected_patch_from_repository is
not going to work, but this one, that I inferred from the old code
should keep things sailing along nicely although I means I need to
amend some patches.
Jason
Tue Aug 19 16:33:48 PDT 2008 Jason Dagit <[EMAIL PROTECTED]>
* Fix a bug with incorrectly specified with_selecte_patch_from_repository.
New patches:
[Fix a bug with incorrectly specified with_selecte_patch_from_repository.
Jason Dagit <[EMAIL PROTECTED]>**20080819233348] hunk ./src/Darcs/Commands/AmendRecord.lhs 120
- with_selected_patch_from_repo "amend" repository opts True $ \ (oldp :> _) -> do
+ with_selected_patch_from_repo "amend" repository opts True $ \ (oldp, _) -> do
hunk ./src/Darcs/SelectChanges.lhs 44
-import Darcs.Patch ( RepoPatch, Patchy, Prim, summary,
+import Darcs.Patch ( RepoPatch, Patchy, Prim, summary, commute,
hunk ./src/Darcs/SelectChanges.lhs 48
-import Darcs.Ordered ( FL(..), RL(..), (:>)(..),
+import Darcs.Ordered ( FL(..), RL(..), (:>)(..), EqCheck(..), MyEq(..),
hunk ./src/Darcs/SelectChanges.lhs 51
- unsafeCoerceP )
+ unsafeCoerceP, reverseRL )
hunk ./src/Darcs/SelectChanges.lhs 193
- -> (FORALL(a) ((PatchInfoAnd p) :> (FL (PatchInfoAnd p))) C(a r) -> IO ()) -> IO ()
+ -> (FORALL(a b) ((PatchInfoAnd p) C(r b), FL (PatchInfoAnd p) C(a r)) -> IO ()) -> IO ()
hunk ./src/Darcs/SelectChanges.lhs 203
- Just (FlippedSeal selected_and_skipped) -> job selected_and_skipped
- Nothing -> do putStrLn $ "Cancelling "++jn++" since no patch was selected."
- exitWith $ ExitSuccess
+ Just (FlippedSeal (s_and_pend :> selected)) ->
+ case reverseFL s_and_pend of
+ pend' :<: skipped ->
+ -- Due to the way wspfr always leaves the pending as the last patch
+ -- in the s_and_pend portion of it's return value, this test should
+ -- always be true and it tells the type system that pend_ and pend'
+ -- are equal and hence have the same contexts.
+ case pend_ `sloppyEquals` pend' of
+ IsEq ->
+ case commute (pend' :> selected) of
+ Just (selected' :> _) -> job (selected', reverseRL skipped)
+ Nothing -> impossible
+ _ -> impossible
+ _ -> impossible
+ Nothing -> impossible
+ where
+ sloppyEquals :: (PatchInfoAnd p) C(x y) -> (PatchInfoAnd p) C(q v) -> EqCheck C(x q)
+ sloppyEquals p1 p2 = p1 =/\= (unsafeCoerceP p2)
hunk ./src/Darcs/SelectChanges.lhs 227
- -> IO (Maybe (FlippedSeal ((PatchInfoAnd p) :> FL (PatchInfoAnd p)) C(z)))
+ -> IO (Maybe (FlippedSeal (FL (PatchInfoAnd p) :> (PatchInfoAnd p)) C(u)))
hunk ./src/Darcs/SelectChanges.lhs 236
- Just _ -> do -- We did this just to be sure that the patches commute freely
+ Just (skipped' :> p') -> do
hunk ./src/Darcs/SelectChanges.lhs 249
- 'y' -> return $ Just $ flipSeal $ p :> skipped
+ 'y' -> return $ Just $ flipSeal $ skipped' :> p'
Context:
[resolve conflicts with Darcs.Ordered change.
David Roundy <[EMAIL PROTECTED]>**20080819153558]
[correct a bug in with_selected_patches_from_repository
Jason Dagit <[EMAIL PROTECTED]>**20080815201125
Fixing this bug allows w_s_p_f_r to have the expected type signature
and also corrects a problem where the non-selected patches were returned
in the wrong context.
]
[fix incompatibility with ghc 6.6.
David Roundy <[EMAIL PROTECTED]>**20080819161540]
[make scan_bundle create a lazy patch bundle.
David Roundy <[EMAIL PROTECTED]>**20080819154624
The idea is to more easily catch errors in which the patch bundle is
misused.
]
[replace separate_middle_last_from_first and separate_last_from_first_middle with get_choices
Jason Dagit <[EMAIL PROTECTED]>**20080815222635]
[refine type witnesses in SelectChanges
Jason Dagit <[EMAIL PROTECTED]>**20080813050425]
[make WhatsNew work with type witnesses
Jason Dagit <[EMAIL PROTECTED]>**20080813044354]
[major refactor of SelectChanges to work with type witnesses
Jason Dagit <[EMAIL PROTECTED]>**20080813031625]
[Eliminate HopefullyPrivate (fixed patch edition)
[EMAIL PROTECTED]
[Move Darcs.Patch.Ordered to Darcs.Ordered since it isn't patchy
[EMAIL PROTECTED]
[Fix use of threadWaitRead on Windows
Simon Marlow <[EMAIL PROTECTED]>**20080819141151
threadWaitRead doesn't work on Windows in all GHC versions < 6.10.1
(which isn't released yet).
This means that since darcs is compiled with -threaded, when compiled
with GHC < 6.10 on Windows, darcs will not respond to ^C when waiting
for user input.
]
[Fix Windows build
Simon Marlow <[EMAIL PROTECTED]>**20080819134252
On Windows, System.Posix.Types.FileOffset is not the same as the type
of the st_size field of the stat structure: the latter is Int64,
whereas COff == Int32.
This is almost ceratinly not the right fix, but it gets the build
going.
In general I don't recommend using System.Posix.* on Windows. The
right way is to either use the official platform-independent libraries
like System.IO, System.Directory or System.Process, or to use
System.Win32 directly.
]
[URL module refactoring.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080817124816]
[Do not compile curl support if libwww is enabled.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080817121834]
[URL.hs: always import debugFail, again.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080817121101]
[URL.hs: call debugFail when download fails.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080816070213]
[URL.hs: debug messages.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080816070147]
[Spacing in Darcs/Commands/Changes.lhs
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080815221245]
[Rename catchInt to catchInterrupt, better message when get is interrupted.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080815154421]
[Resolve issue995: changes --context : {--human-readable,--xml-output} have no effect.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080815114105]
[Tidy up known failing shell tests.
Eric Kow <[EMAIL PROTECTED]>**20080818102100
* Keeping track of these bugs could be simpler if they were all
associated with a ticket and a short description
* The shell harness allows us to avoid a lot of bureaucracy in
in the scripts. In fact, people should be able to submit a
bug just by sticking a shell script in bugs, no hoops to jump.
]
[Tidy up failing test for issue1013.
Eric Kow <[EMAIL PROTECTED]>**20080818101351
We don't need to use a $DARCS variable anymore, thanks to the shell
harness. Also, I noticed that what used to trigger a bug now hangs.
]
[Add failing test for issue1012 (Simon Marlow).
Eric Kow <[EMAIL PROTECTED]>**20080818094104]
[Add test for issue1017 (as suggested by Bjorn Bringert)
Eric Kow <[EMAIL PROTECTED]>**20080818194115]
[don't print "Wrote patch to -" when writing patch to stdout.
David Roundy <[EMAIL PROTECTED]>**20080819142537]
[always send with context.
David Roundy <[EMAIL PROTECTED]>**20080819140729]
[Resolve issue823: do not exit on keyboard interrupt when getting patches.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080815070943
And give a chance for go_to_chosen_version to run.
]
[fix buggy comments in bugs/identical-patches.sh.
David Roundy <[EMAIL PROTECTED]>**20080814135322]
[Add Ian's identical-patch test case.
Eric Kow <[EMAIL PROTECTED]>**20080813171032]
[URL.hs: store only URL in waitToStart queue.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080813122246]
[Add (failing) test for issue944.
Eric Kow <[EMAIL PROTECTED]>**20080814055903
This appears to be a reggression from darcs 1.0.9, and was submitted by
Wolfgang Lux on the bug tracker. Interestingly, only the old format
repositories are affected, not the hashed ones.
]
[add type witnesses to TouchesFiles
Jason Dagit <[EMAIL PROTECTED]>**20080810063403]
[add type witnesses to Patch/Choices.lhs
Jason Dagit <[EMAIL PROTECTED]>**20080809000237]
[Split Cache mostly out of Darsc/Repository/Prefs into its own file (take 2)
[EMAIL PROTECTED]
[Make Darcs.Repository.Prefs export the cache hash function
[EMAIL PROTECTED]
[remove a few unsightly functions
Jason Dagit <[EMAIL PROTECTED]>**20080813061256]
[Fix URL module bug with pipelining enabled.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080813081218]
[Minor change to URL module.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080813074218]
[Enable pipelining by default, add --disable-pipelining option (issue838).
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080813011342]
[Generalize HashRepo.clean_pristine to HashIO.clean_hashdir.
[EMAIL PROTECTED]
[Add writeSlurpy to roll out a copy of slurpy into a filesystem.
[EMAIL PROTECTED]
[fix breakage in URL.
David Roundy <[EMAIL PROTECTED]>**20080812141220]
[Parametrize "pristine.hashed" in a bunch of functions.
[EMAIL PROTECTED]
[Rework URL module for multi threading.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080811221209]
[Add thread synchronization to URL module and resume select() if interrupted by signal in curl module.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080810092810]
[Handle error case with empty URL in URL.waitNextUrl function.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080809221755]
[Add --debug-http flag to enable curl and libwww debug at run-time instead of compile-time.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080809154834]
[Print a warning when the remote end does not have darcs 2.
Eric Kow <[EMAIL PROTECTED]>**20080811100933
Two reasons:
(1) right now people get a scary warning from ssh when it can't fetch
some non-essential files (it used to be that we would send stderr from ssh
to /dev/null, but that has other problems...)
(2) darcs transfer-mode more widely deployed could help a lot of people
wrt darcs performance
]
[Added a beware note to the unrecord command
[EMAIL PROTECTED]
[Fixed typo
[EMAIL PROTECTED]
[Better debug messages in URL module.
Dmitry Kurochkin <[EMAIL PROTECTED]>**20080809215247]
[make Convert.lhs compile.
David Roundy <[EMAIL PROTECTED]>**20080810201725]
[improve type safety of Darcs.Repository.Internal.
Jason Dagit <[EMAIL PROTECTED]>**20080810051109]
[Refactor `darcs convert' warning at kowey's request.
Trent W. Buck <[EMAIL PROTECTED]>**20080810110014]
[Expand formats text based in part on suggestions from darcs-users
Max Battcher <[EMAIL PROTECTED]>**20080809184043]
[Fixes to global cache text based on darcs-users suggestions
Max Battcher <[EMAIL PROTECTED]>**20080809181424]
[Add user-focused documentation of repository format options
Max Battcher <[EMAIL PROTECTED]>**20080807195429]
[Highlight the global cache as a best practice
Max Battcher <[EMAIL PROTECTED]>**20080807193918]
[Describe best practice in `darcs convert --help'.
Trent W. Buck <[EMAIL PROTECTED]>**20080810110615]
[add type witnesses to Population
Jason Dagit <[EMAIL PROTECTED]>**20080808053252]
[add type witnesses to CommandsAux
Jason Dagit <[EMAIL PROTECTED]>**20080808052738]
[Add type witnesses to more modules, rounding out Darcs/Repository/*
Jason Dagit <[EMAIL PROTECTED]>**20080808050947]
[fixed a bug in identity_commutes property
Jason Dagit <[EMAIL PROTECTED]>**20080808023025
In the right identity check the patch order should have gone from
(identity :> p) to (p2 :> i2). I added a rigid type context too
so that ghc 6.8 and newer would type the definition.
]
[Make Darcs.Repository.Internal compile with type witnesses.
Jason Dagit <[EMAIL PROTECTED]>**20080808015343]
[UF8.lhs: remove unusued functions/imports/docs
[EMAIL PROTECTED]
[Resolve issue974 : do not pass both -optc-g and -opta-g to GHC
Eric Kow <[EMAIL PROTECTED]>**20080807073620]
[make this test more cross-platform
Simon Michael <[EMAIL PROTECTED]>**20080807103433]
[document how to run unit tests
Simon Michael <[EMAIL PROTECTED]>**20080807030416]
[move (most) failing tests to bugs for clean test output
Simon Michael <[EMAIL PROTECTED]>**20080806191336]
[fix an old spelling error
Simon Michael <[EMAIL PROTECTED]>**20080806170432]
[make searching for "test:" in makefile work
Simon Michael <[EMAIL PROTECTED]>**20080805222241]
[run only normal (expected to pass) tests by default
Simon Michael <[EMAIL PROTECTED]>**20080805222108]
[Downplay quantum mechanics link.
Eric Kow <[EMAIL PROTECTED]>**20080806124109
Besides, darcs has far more than 3 users by now.
]
[Make patch theory intro more inviting to math people.
Eric Kow <[EMAIL PROTECTED]>**20080806123411]
[cleanup and slight rewrite of the test docs
Simon Michael <[EMAIL PROTECTED]>**20080806165949]
[make order of running tests consistent
Simon Michael <[EMAIL PROTECTED]>**20080806172123]
[small makefile refactoring: allow just the normal tests to be run, without bugs/*
Simon Michael <[EMAIL PROTECTED]>**20080805203242]
[Rectify dist help
[EMAIL PROTECTED]
Removed the "make dist" suggestion, the manual is a better place for that.
Instead, make clear that it operates on a clean copy of the tree, and
mention the "predist" functionality.
]
[website: explain that darcs 2 is required to get the darcs source.
Simon Michael <[EMAIL PROTECTED]>**20080803181216]
[Canonize Gaetan Lehmann and Daniel Buenzli.
Eric Kow <[EMAIL PROTECTED]>**20080730104357
(for Daniel B, avoid an accent in his name)
]
[configure: check for packages needed with split base.
Eric Kow <[EMAIL PROTECTED]>**20080730103840
Now that all packages must be used explicitly.
]
[fix type witness compile errors specific to ghc 6.8
Jason Dagit <[EMAIL PROTECTED]>**20080722182729]
[avoid import of unused function fromMaybe.
David Roundy <[EMAIL PROTECTED]>**20080729172825]
[configure: suggest regex-compat before text
Eric Kow <[EMAIL PROTECTED]>**20080725095336]
[configure: mention Haskell in 'try installing' suggestion
Eric Kow <[EMAIL PROTECTED]>**20080725095015]
[Typo (Text.Regex)
Eric Kow <[EMAIL PROTECTED]>**20080715121708]
[Use haskeline to have a readline-like behavior when asking something to the user
[EMAIL PROTECTED]
Unlike the implementations using readline or editline packages, this code
code doesn't break the Ctrl-C behavior.
]
[Improve generic rules for English plurals.
Eric Kow <[EMAIL PROTECTED]>**20080604123728]
[add configure check for Network.URI.
David Roundy <[EMAIL PROTECTED]>**20080711011914]
[add -hide-all-packages to default GHCFLAGS.
David Roundy <[EMAIL PROTECTED]>**20080711010952]
[add support for outputting patch numbers in darcs changes.
David Roundy <[EMAIL PROTECTED]>**20080710011211]
[add support for matching single patches by index.
David Roundy <[EMAIL PROTECTED]>**20080710004512]
[add support for matching ranges of patches (counting back from present).
David Roundy <[EMAIL PROTECTED]>**20080710003225]
[Better avoid silly manpage error.
Trent W. Buck <[EMAIL PROTECTED]>**20080704024920
It turned out only initialize's help string used 'quotes', so just
remove them. This makes init's docstring consistent with the others.
]
[Missing period at end of sentence.
Trent W. Buck <[EMAIL PROTECTED]>**20080704024232]
[darcs --overview no longer works, so don't document it.
Trent W. Buck <[EMAIL PROTECTED]>**20080704030804]
[Avoid silly manpage error.
Trent W. Buck <[EMAIL PROTECTED]>**20080703010733
man (nroff) treats an apostrophe in the first column specially,
resulting in a syntax error without this patch.
Ideally, all cases of 'foo' in the manpage (i.e. docstrings) should
become `foo', since man -Tps turns ` and ' into left and right single
quotes respectively.
]
[obliterate whitespace in Darcs.Commands.Get
[EMAIL PROTECTED]
'twas causing lhs/haddock difficulties where a \end{code} wasn't getting recognized.
]
[rm haddock CPP business
[EMAIL PROTECTED]
Try as I might, I can't see any reason to special-case some Haddock CPP logic to deal with some *commented-out guards*, unless CPP magically restores and uncomments the code if Haddock isn't being run.
]
[make pull less verbose when --verbose flag is given.
David Roundy <[EMAIL PROTECTED]>**20080624170035]
[fix makefile to remember to regenerate version information after running configure.
David Roundy <[EMAIL PROTECTED]>**20080624170001]
[TAG 2.0.2
David Roundy <[EMAIL PROTECTED]>**20080624012041]
Patch bundle hash:
6fc55d7b8573bbfa42becd788a2031e7ed5f307c
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users
