CEDET

Eric M. Ludlam wrote:
>>>> <klaus.berndl@...> seems to think that:
>> klaus.berndl@... wrote:
>   [ ... ]
>>>>   In the above example, you would find that
>>>>=20
>>>> M-x semantic-analyze-current-context
>>>>=20
>>>>   can return some of the right stuff, but only if the point is in
>>>> the body of the method (that you would have to add.)
>>>=20
>>> When i call this command then i always get only the message:
>>> "semantic-up-context: No context of type function to advance in [4
>>> times]"
>=20
> Yes, you have to type in a method with a body, put the cursor in the
> body, and execute the above for it to work.
>=20
>>>>=20
>>>>   The magic function semantic-ctxt-scoped-types also isn't quite
>>>> good enough because it assumes that all namespaces in the local
>>>> file are in scope!=20
>>>>=20
>>>>   I suppose an ideal solution might be to fix the analyze-context
>>>> function (or something like it) to put the active scopes into some
>>>> sort of specialized order, then search them for your symbol.
>>>=20
>>> Why not to enable the magic semantic-ctxt-scoped-types to take into
>>> account Scope, means in the example above, that when called from
>>> within namespace moose It returns only the moose-table and not the
>>> meese-table?!=20
>>>=20
>>> Is this "...it assumes that all namespaces in the local file are in
>>> scope!" a senseful behavior or is assumption a good assumption? -
>>> why it assumes It?
>   [ ... ]
>=20
> That is the default for C/C++.  The default returns nothing at all.
>=20
> I would be happy if anyone who really knows C++ would look at the
> implementation in semantic-c.by

Where you would say we should take a look - this files is quite big? ;-)

>and the method
> semantic-ctxt-scoped-types and make it "right."  I'm not familiar
> enough with C++ to divine the right answer myself.

Hmm, i can make a try..... Do i understand you right, that =
semantic-ctxt-scoped-types
should be the producer of the right types currently in scope?!

Ciao,
Klaus

Re[3]: FW: AW: [CEDET-devel] [ECB-list] Parent classes aren't jump-able From: Eric M. Ludlam <eric@si...> - 2005-06-29 09:37 >>> <klaus.berndl@...> seems to think that: >Eric M. Ludlam wrote: >>>>> <klaus.berndl@...> seems to think that: >>> klaus.berndl@... wrote: >> [ ... ] >>>>> In the above example, you would find that >>>>> >>>>> M-x semantic-analyze-current-context >>>>> >>>>> can return some of the right stuff, but only if the point is in >>>>> the body of the method (that you would have to add.) >>>> >>>> When i call this command then i always get only the message: >>>> "semantic-up-context: No context of type function to advance in [4 >>>> times]" >> >> Yes, you have to type in a method with a body, put the cursor in the >> body, and execute the above for it to work. >> >>>>> >>>>> The magic function semantic-ctxt-scoped-types also isn't quite >>>>> good enough because it assumes that all namespaces in the local >>>>> file are in scope! >>>>> >>>>> I suppose an ideal solution might be to fix the analyze-context >>>>> function (or something like it) to put the active scopes into some >>>>> sort of specialized order, then search them for your symbol. >>>> >>>> Why not to enable the magic semantic-ctxt-scoped-types to take into >>>> account Scope, means in the example above, that when called from >>>> within namespace moose It returns only the moose-table and not the >>>> meese-table?! >>>> >>>> Is this "...it assumes that all namespaces in the local file are in >>>> scope!" a senseful behavior or is assumption a good assumption? - >>>> why it assumes It? >> [ ... ] >> >> That is the default for C/C++. The default returns nothing at all. >> >> I would be happy if anyone who really knows C++ would look at the >> implementation in semantic-c.by > >Where you would say we should take a look - this files is quite big? ;-) Uh, sorry, it is semantic-c.el. In there, just search for the below function. >>and the method >> semantic-ctxt-scoped-types and make it "right." I'm not familiar >> enough with C++ to divine the right answer myself. > >Hmm, i can make a try..... Do i understand you right, that semantic-ctxt-scoped-types >should be the producer of the right types currently in scope?! [ ... ] Right. For a given location (point), it should return a list of tags which represent symbols that are in scope. ie, if the symbol 'foo' is near the cursor, then one of moose::foo, or meese::foo could be in scope and could be substituted in for the plain symbol foo. Eric -- Eric Ludlam: zappo@..., eric@... Home: http://www.ludlam.net Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net GNU: http://www.gnu.org

RE: Re[3]: FW: AW: [CEDET-devel] [ECB-list] Parent classes aren't jump-able From: <klaus.berndl@sd...> - 2005-06-29 15:47 >>> That is the default for C/C++. The default returns nothing at all. >>>=20 >>> I would be happy if anyone who really knows C++ would look at the >>> implementation in semantic-c.by >>=20 >> Where you would say we should take a look - this files is quite big? >> ;-)=20 >=20 > Uh, sorry, it is semantic-c.el. Ah, now i can follow you much more ;-) > In there, just search for the below function. >=20 >>> and the method >>> semantic-ctxt-scoped-types and make it "right." I'm not familiar >>> enough with C++ to divine the right answer myself. >>=20 >> Hmm, i can make a try..... Do i understand you right, that >> semantic-ctxt-scoped-types should be the producer of the right types >> currently in scope?!=20 > [ ... ] >=20 > Right. For a given location (point), it should return a list of tags > which represent symbols that are in scope. ie, if the symbol 'foo' is > near the cursor, then one of moose::foo, or meese::foo could be in > scope and could be substituted in for the plain symbol foo. Hmm,lets go back to this example: namespace moose { =20 class fee { int fi; }; =20 class fie : public fee { int fi; }; =20 } namespace meese { =20 class fee { int fi; }; =20 class fey : fee { int fi; =20 void print(); ^ <---------------- <point> (B) }; =20 void Meese::print() { ^ <---------------- <point> (A) } } Suppose, point is at point(A) like above illustrated...Then=20 semantic-ctxt-scoped-types should return what )currently it returns Both tag-tables moose and meese which is "nonsense" ;-) Then it should return the tag-table meese. When point is at (B) then it should return what? Also tag-tabel meese? I currently do not know exactly the intention of = semantic-ctxt-scoped-types: Sould it return always which types are available in scope so i can use = it at current point?? But what about types not defined in current file?? = Suppose I write "using anotherNamespace;" at the top of this example. Then all = types of anotherNamespace should be also in scope and = semantic-ctxt-scoped-types Should return it (but for this it would require semanticdb!)... Hmm, have i described this understandable or are you now confused more = than before? ;-) Ciao, Klaus >=20 > Eric

1 message has been excluded from this view by a project administrator.