API docs for ThreadedCompleter

This commit is contained in:
Strahinja Val Markovic 2013-05-13 13:55:02 -07:00
parent 381b86d595
commit 4389b1f1d8
2 changed files with 25 additions and 2 deletions

View File

@ -64,7 +64,8 @@ class Completer( object ):
CandidatesForQueryAsync() is the main entry point when the user types. For
"foo.bar", the user query is "bar" and completions matching this string should
be shown. The job of CandidatesForQueryAsync() is to merely initiate this
request, which will hopefully be processed in a background thread.
request, which will hopefully be processed in a background thread. You may
want to subclass ThreadedCompleter instead of Completer directly.
AsyncCandidateRequestReady() is the function that is repeatedly polled until
it returns True. If CandidatesForQueryAsync() started a background task of
@ -94,7 +95,10 @@ class Completer( object ):
thread-safe). But even if you're certain, still try to do the processing in a
background thread. Your completer is unlikely to be merged if it does not,
because synchronous processing will block Vim's GUI thread and that's a very,
VERY bad thing (so try not to do it!).
VERY bad thing (so try not to do it!). Again, you may want to subclass
ThreadedCompleter instead of Completer directly; ThreadedCompleter will
abstract away the use of a background thread for you. See
threaded_completer.py.
The On* functions are provided for your convenience. They are called when
their specific events occur. For instance, the identifier completer collects

View File

@ -22,6 +22,22 @@ from threading import Thread, Event
from ycm.completers.completer import Completer
class ThreadedCompleter( Completer ):
"""A subclass of Completer that abstracts away the use of a background thread.
This is a great class to subclass for your custom completer. It will provide
you with async computation of candidates when you implement the
ComputeCandidates() method; no need to worry about threads, locks, events or
similar.
If you use this class as the base class for your Completer, you DON'T need to
provide implementations for the CandidatesForQueryAsync(),
AsyncCandidateRequestReady() and CandidatesFromStoredRequest() functions. Just
implement ComputeCandidates().
For examples of subclasses of this class, see the following files:
python/completers/general/filename_completer.py
"""
def __init__( self ):
super( ThreadedCompleter, self ).__init__()
self._query_ready = Event()
@ -54,6 +70,9 @@ class ThreadedCompleter( Completer ):
@abc.abstractmethod
def ComputeCandidates( self, query, start_column ):
"""This function should compute the candidates to show to the user.
The return value should be of the same type as that for
CandidatesFromStoredRequest()."""
pass