Addresses review comments from https://git.eclipse.org/r/#/c/10648.
Fixes the junit problems by making sure that the dummy PDOM acquires its
write lock before calling exercising the tag index.
Original commit message:
This new extension point allows contributors to put their own
information into the PDOM and to later retrieve it for their own
purposes.
There are many details in the bug. The idea is that contributors
provide an implementation of IBindingTagger, which is given a chance to
examine IBindings when they are created. The ITagWriter interface
allows the contributor to create a new tag which can then have data
written to it.
The ITagService interface (accessible from CCorePlugin.getTagService()
provides a way for the contributor to later get an instance of
ITagReader to retrieve tags from bindings.
ITags are copied to the PDOM when the associated binding is persisteed.
Contributors use a unique id (based on their plugin id), so that
multiple contributors are able to independently tag a given binding.
In-memory tags are not cached. I've done some timing tests using my
sample implementation and found no measurable difference. The full log
lines look like:
!MESSAGE Indexed 'simple-01' (2 sources, 184 headers) in <see below>
sec: 21,550 declarations; 35,394 references; 0 unresolved inclusions; 1
syntax errors; 0 unresolved names (0.00%)
I did 5 tests using the current master (no tagging-related code), the
times were:
18.86 sec
9.17 sec
5.91 sec
4.79 sec
4.83 sec
And then I ran the same sequence of tests using the code in this
commit:
18.73 sec
9.39 sec
6.50 sec
4.78 sec
5.27 sec
If performance does become a problem, then caching could be introduced
with a new implementation of ITaggableService. The two problems are
finding a key other than the identity of the IBinding (since IBindings
are re-created often) and properly evicting stale entries when the
binding is no longer valid.
The process of copying tags from an in-memory IBinding to a PDOMBinding,
is a synchronization. This means that tags that are no longer
applicable, will be removed from the persistent store.
While developing this I found that PDOMBindings are not deleted from the
Database (only the names that reference them are deleted), so there is
no provision for deleting all tags at once.
New database locks are not needed. By the time the persistent tags are
accessed, higher levels of code have already taken a read or write lock
as appropriate.
There are new unit tests covering the changes to the PDOM.
Change-Id: I6ae1afc949082f7f4484b3faa1550670be43312f
Reviewed-on: https://git.eclipse.org/r/10659
Reviewed-by: Doug Schaefer <dschaefer@qnx.com>
IP-Clean: Doug Schaefer <dschaefer@qnx.com>
Tested-by: Doug Schaefer <dschaefer@qnx.com>
Some of the descriptions are not very good, and will be improved in a
future commit.
Change-Id: I1150a76c504fe0cf71911acb3c0a1c6e72ee3686
Reviewed-on: https://git.eclipse.org/r/10557
Reviewed-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
IP-Clean: Sergey Prigogin <eclipse.sprigogin@gmail.com>
Tested-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
The edit menu docs will be updated in a later patch.
Change-Id: I5a8975d1159e39e580e87f502160d3f46d71b3f0
Reviewed-on: https://git.eclipse.org/r/10604
Reviewed-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
IP-Clean: Sergey Prigogin <eclipse.sprigogin@gmail.com>
Tested-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
directory should be disallowed
Change-Id: I75090498bd4f408d77ad17491f50f4fd66850c97
Reviewed-on: https://git.eclipse.org/r/10404
Reviewed-by: Jesse Weinstein <Jesse.Weinstein@clinicomp.com>
Reviewed-by: Jeff Johnston <jjohnstn@redhat.com>
IP-Clean: Jeff Johnston <jjohnstn@redhat.com>
Tested-by: Jeff Johnston <jjohnstn@redhat.com>
Tags signal and slot methods when the index is created. Uses these tags
to suggest values inside of SIGNAL and SLOT macro expansions. Enabled
only for projects with the QtNature.
Recognizes QObject::connect function calls and suggests SIGNAL(a) and
SLOT(a) for the 2nd and 4th parameters.
When expanding the SIGNAL and SLOT macros within a call to
QObject::connect, suggests signals and slots based on the type of the
previous parameter.
E.g. in
QObjectA a;
QObjectB b;
QObject::connect( &a, SIGNAL(*), &b, SLOT(**) );
The content assistant will suggest the methods of type QObjectA that
have been marked as signals at *, and the methods of QObjectB that have
been marked as slots at **.
Change-Id: Ia6aaa71724547b0977e322399a500f072004767a
Reviewed-on: https://git.eclipse.org/r/10532
Reviewed-by: Doug Schaefer <dschaefer@qnx.com>
IP-Clean: Doug Schaefer <dschaefer@qnx.com>
Tested-by: Doug Schaefer <dschaefer@qnx.com>
This new extension point allows contributors to put their own
information into the PDOM and to later retrieve it for their own
purposes.
There are many details in the bug. The idea is that contributors
provide an implementation of IBindingTagger, which is given a chance to
examine IBindings when they are created. The ITagWriter interface
allows the contributor to create a new tag which can then have data
written to it.
The ITagService interface (accessible from CCorePlugin.getTagService()
provides a way for the contributor to later get an instance of
ITagReader to retrieve tags from bindings.
ITags are copied to the PDOM when the associated binding is persisteed.
Contributors use a unique id (based on their plugin id), so that
multiple contributors are able to independently tag a given binding.
In-memory tags are not cached. I've done some timing tests using my
sample implementation and found no measurable difference. The full log
lines look like:
!MESSAGE Indexed 'simple-01' (2 sources, 184 headers) in <see below>
sec: 21,550 declarations; 35,394 references; 0 unresolved inclusions; 1
syntax errors; 0 unresolved names (0.00%)
I did 5 tests using the current master (no tagging-related code), the
times were:
18.86 sec
9.17 sec
5.91 sec
4.79 sec
4.83 sec
And then I ran the same sequence of tests using the code in this
commit:
18.73 sec
9.39 sec
6.50 sec
4.78 sec
5.27 sec
If performance does become a problem, then caching could be introduced
with a new implementation of ITaggableService. The two problems are
finding a key other than the identity of the IBinding (since IBindings
are re-created often) and properly evicting stale entries when the
binding is no longer valid.
The process of copying tags from an in-memory IBinding to a PDOMBinding,
is a synchronization. This means that tags that are no longer
applicable, will be removed from the persistent store.
While developing this I found that PDOMBindings are not deleted from the
Database (only the names that reference them are deleted), so there is
no provision for deleting all tags at once.
New database locks are not needed. By the time the persistent tags are
accessed, higher levels of code have already taken a read or write lock
as appropriate.
There are new unit tests covering the changes to the PDOM.
Change-Id: I8da1bf5eeba7e1fc2ca7ec308ed8e212629986a4
Reviewed-on: https://git.eclipse.org/r/10407
IP-Clean: Doug Schaefer <dschaefer@qnx.com>
Tested-by: Doug Schaefer <dschaefer@qnx.com>
Reviewed-by: Doug Schaefer <dschaefer@qnx.com>
first parameter in constructor
createBinding is called twice for the enumeration specifier because now
ImplicitsAnalysis calls createType on the constructor parameter
declaration. The second time it is called, it wrongly thinks the
specifier is a redefinition. We can check that it's not a redefinition
by comparing the ast names.
Change-Id: If7fb168d578a7a59e78867d6d8154e0866d2a730
Reviewed-on: https://git.eclipse.org/r/10272
Reviewed-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
IP-Clean: Sergey Prigogin <eclipse.sprigogin@gmail.com>
Tested-by: Sergey Prigogin <eclipse.sprigogin@gmail.com>
language settings providers are enabled
Change-Id: Ie9cbb2df8dbc9b1034bde2dbfd5de64d27d54c49
Reviewed-on: https://git.eclipse.org/r/10248
Reviewed-by: Andrew Gvozdev <angvoz.dev@gmail.com>
IP-Clean: Andrew Gvozdev <angvoz.dev@gmail.com>
Tested-by: Andrew Gvozdev <angvoz.dev@gmail.com>
Change-Id: I09ce5046ccc0935ef036bb6a707900a308f7204f
Reviewed-on: https://git.eclipse.org/r/10255
Reviewed-by: William Swanson <traveler@tilera.com>
Reviewed-by: Marc Khouzam <marc.khouzam@ericsson.com>
IP-Clean: Marc Khouzam <marc.khouzam@ericsson.com>
Tested-by: Marc Khouzam <marc.khouzam@ericsson.com>
