Friendly Interface for auto-completion of GHCi

Now the GHCi can do TAB completion when running in term.

But in Emacs/Inf-Haskell mode, TAB completion doesn't work. It should be great that if GHCi support the command like "complete" in GDB.

If GHCi supports this, it's easy to implement TAB completion in Emacs/Inf-Haskell mode.

Trac metadata

Trac field	Value
Version	7.2.1
Type	FeatureRequest
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	GHCi
Test case
Differential revisions
BlockedBy
Related
Blocking
CC
Operating system
Architecture

changed weight to 5

added GHCi Tfeature request Trac import labels

changed milestone to %7.6.1

changed milestone to %7.6.2

Attached file ticket-5687-ghci-complete.diff (download).

First draft for a :complete command

The attached patch currently exposes the ghciCompleteWord function and uses Read/Show quoted strings for input/output. The first two lines returned are the number of completion candidates and the unused line part (as returned by ghciCompleteWord). The remaining lines are all completions found (up to the given limit).

Here's an example:

Prelude> :complete 
Syntax: :complete repl <max-completions> <quoted-string-to-complete>
Prelude> :complete repl 5 "import For"
21
"import "
"Foreign"
"Foreign.C"
"Foreign.C.Error"
"Foreign.C.String"
"Foreign.C.Types"
Prelude> :complete repl 5 ":"
52
""
":?"
":abandon"
":add"
":back"
":break"
Prelude> :complete repl 5 "map"
3
""
"map"
"mapM"
"mapM_"
Prelude> :complete repl 0 "map"
3
""
Prelude> :complete repl 0 ""
470
""
Prelude> :complete repl -1 "map"
3
""
"map"
"mapM"
"mapM_"

Attached file ticket-5687-ghci-complete.v2.diff (download).

2nd draft for :complete command

I've attached a patch providing a more refined :complete command:

This one takes an optional range (1-based indexing, syntax [n-][m] where n denotes first index, and m the last index shown), allowing to slice the result list; moreover the first line emitted now contains:

<completions-printed-count> <total-completions-count> <line-prefix>

And here's an example session again:

Prelude> :complete 
Syntax: :complete repl [<range>] <quoted-string-to-complete>
Prelude> :complete repl 5 "import For"
5 21 "import "
"Foreign"
"Foreign.C"
"Foreign.C.Error"
"Foreign.C.String"
"Foreign.C.Types"
Prelude> :complete repl 5-10 "import For"
6 21 "import "
"Foreign.C.Types"
"Foreign.Concurrent"
"Foreign.ForeignPtr"
"Foreign.ForeignPtr.Safe"
"Foreign.ForeignPtr.Unsafe"
"Foreign.Marshal"
Prelude> :complete repl 20- "import For"
2 21 "import "
"Foreign.StablePtr"
"Foreign.Storable"
Prelude> :complete repl "map"
3 3 ""
"map"
"mapM"
"mapM_"

Is there any sort of standard for what the output should look like?

hvr says he knows of no standard. He'll add some docs for the user guide, and is also working on an implementation of the emacs side.

hvr: I wonder if it would be better to use your own String-showing function rather than using Show, so as to make it possible to correctly use simpler parsing on the emacs side.

@igloo:

I've got a nicely working proof of concept implementation for Emacs (see http://youtu.be/06q-sCoqwEY for a short screencast demonstration).

Implementing complete Read/Show serialization for strings wasn't that hard in Elisp, and the standard case, when no escaping is needed has a cheap code path.

Attached file ticket-5687-ghci-complete-docs.v2.diff (download).

User Guide entry for :complete comment

mentioned in commit 36a54173

mentioned in commit 195626fc

closed

Applied, thanks

Trac metadata

Trac field	Value
Resolution	Unresolved → ResolvedFixed

added Pnormal label