Home » Language IDEs » C / C++ IDE (CDT) » 'intellisense'
| 'intellisense' [message #80558] |
Fri, 19 September 2003 17:01  |
Eclipse User |
|
|
|
Originally posted by: symme7ry.MICROSOFTSfreeEMAILservice.com
I've seen previous questions asking about 'code assist' in general, but
not specifically about 'intellisense'.
When I type "string s; s.", I'd like a bunch of method signatures to pop
up after I type the period. When I type "getline(", I'd like the method
signature to pop up after I type the parenthesis. Does this particular
feature exist now? Or anything similar? I don't really care that if I type
"f" then control space enter, it fills in a for loop for me. That kind of
code assist I dont care for. I just need some sort of built-in knowledge
of the c++ library. If this does not yet exist, does anyone know when it
will?
I have eclpise version 2.1.1 and cdt version 1.2.0.51
|
|
|
| Re: 'intellisense' [message #80616 is a reply to message #80558] |
Mon, 22 September 2003 09:40  |
| Eclipse User |
|
|
|
Completing after a "." requires parsing the current document and knowing
the type of the variable before the ".". This is not available now for
Code Assist (release CDT 1.2, October 2003); however, it will be available
on the next release (CDT 2.0).
In this release, inside the body of a function or a method, you are able
to find completion proposals of global functions, variables, enums,
macros, etc...
Code assist is using the search feature to find these proposals. Anything
that is globally defined that starts with the given prefix will be
searched for and added as a completion proposal. This requires two things:
1- that you have the new indexer enabled ( it will be enabled by default
by the release date, but right now you have to check the "C/C++ Project"
tab for your project's properties. )
2- that your code is parsing without errors and that your code is saved.
If you have some parsing errors, you do not garantee to find all possible
proposals.
symme7ry wrote:
> I've seen previous questions asking about 'code assist' in general, but
> not specifically about 'intellisense'.
> When I type "string s; s.", I'd like a bunch of method signatures to pop
> up after I type the period. When I type "getline(", I'd like the method
> signature to pop up after I type the parenthesis. Does this particular
> feature exist now? Or anything similar? I don't really care that if I type
> "f" then control space enter, it fills in a for loop for me. That kind of
> code assist I dont care for. I just need some sort of built-in knowledge
> of the c++ library. If this does not yet exist, does anyone know when it
> will?
> I have eclpise version 2.1.1 and cdt version 1.2.0.51
|
|
|
| Re: 'intellisense' [message #80706 is a reply to message #80616] |
Tue, 23 September 2003 08:54 |
| Eclipse User |
|
|
|
This is a real pity as it means that our company will not be able to use
the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse 3.0 it
will be more then a year, all this to get a feature that are considered
minimum for developers. Is there an alternative?
The ast parser is already running, so it shouldn't be too hard to find out
the type or am I completely wrong here?
Anyway apart from this the CDT has really made some great steps forward.
LE
> Completing after a "." requires parsing the current document and knowing
> the type of the variable before the ".". This is not available now for
> Code Assist (release CDT 1.2, October 2003); however, it will be available
> on the next release (CDT 2.0).
> In this release, inside the body of a function or a method, you are able
> to find completion proposals of global functions, variables, enums,
> macros, etc...
> Code assist is using the search feature to find these proposals. Anything
> that is globally defined that starts with the given prefix will be
> searched for and added as a completion proposal. This requires two things:
> 1- that you have the new indexer enabled ( it will be enabled by default
> by the release date, but right now you have to check the "C/C++ Project"
> tab for your project's properties. )
> 2- that your code is parsing without errors and that your code is saved.
> If you have some parsing errors, you do not garantee to find all possible
> proposals.
> symme7ry wrote:
> > I've seen previous questions asking about 'code assist' in general, but
> > not specifically about 'intellisense'.
> > When I type "string s; s.", I'd like a bunch of method signatures to pop
> > up after I type the period. When I type "getline(", I'd like the method
> > signature to pop up after I type the parenthesis. Does this particular
> > feature exist now? Or anything similar? I don't really care that if I type
> > "f" then control space enter, it fills in a for loop for me. That kind of
> > code assist I dont care for. I just need some sort of built-in knowledge
> > of the c++ library. If this does not yet exist, does anyone know when it
> > will?
> > I have eclpise version 2.1.1 and cdt version 1.2.0.51
|
|
|
| Re: 'intellisense' [message #80752 is a reply to message #80706] |
Tue, 23 September 2003 11:29 |
| Eclipse User |
|
|
|
The CDT is toolchain agnostic, which means that the parser is expected to
work for most specified compilers. Even if the compilers are ANSI compliant,
there can still be problems.
There is a huge amount of verification we need to do to get things like
STDIO.H parsing on different platforms, due to the amount of internal
compiler macro definitions. If you use gcc, for example, and do 'gcc -v',
you'll see it reads in its specs from some text file in your install that
tells the compiler what extra #defines to add and re-order the include path.
Getting this to work for Standard Make projects is very difficult given
these requirements, as we have no understanding of what toolchain is
provided. We are working quite diligently to provide solutions in the
future that will provide code assist/search/refactoring features as good as
the JDT.
JohnC
"Eli" <elil@amdocs.com> wrote in message news:bkpfqf$306$1@eclipse.org...
> This is a real pity as it means that our company will not be able to use
> the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse 3.0 it
> will be more then a year, all this to get a feature that are considered
> minimum for developers. Is there an alternative?
>
> The ast parser is already running, so it shouldn't be too hard to find out
> the type or am I completely wrong here?
>
> Anyway apart from this the CDT has really made some great steps forward.
>
> LE
>
>
> > Completing after a "." requires parsing the current document and knowing
> > the type of the variable before the ".". This is not available now for
> > Code Assist (release CDT 1.2, October 2003); however, it will be
available
> > on the next release (CDT 2.0).
> > In this release, inside the body of a function or a method, you are able
> > to find completion proposals of global functions, variables, enums,
> > macros, etc...
> > Code assist is using the search feature to find these proposals.
Anything
> > that is globally defined that starts with the given prefix will be
> > searched for and added as a completion proposal. This requires two
things:
> > 1- that you have the new indexer enabled ( it will be enabled by default
> > by the release date, but right now you have to check the "C/C++ Project"
> > tab for your project's properties. )
> > 2- that your code is parsing without errors and that your code is saved.
> > If you have some parsing errors, you do not garantee to find all
possible
> > proposals.
>
> > symme7ry wrote:
>
> > > I've seen previous questions asking about 'code assist' in general,
but
> > > not specifically about 'intellisense'.
>
> > > When I type "string s; s.", I'd like a bunch of method signatures to
pop
> > > up after I type the period. When I type "getline(", I'd like the
method
> > > signature to pop up after I type the parenthesis. Does this particular
> > > feature exist now? Or anything similar? I don't really care that if I
type
> > > "f" then control space enter, it fills in a for loop for me. That kind
of
> > > code assist I dont care for. I just need some sort of built-in
knowledge
> > > of the c++ library. If this does not yet exist, does anyone know when
it
> > > will?
>
> > > I have eclpise version 2.1.1 and cdt version 1.2.0.51
>
>
|
|
|
| Re: 'intellisense' [message #80938 is a reply to message #80752] |
Wed, 24 September 2003 02:31 |
| Eclipse User |
|
|
|
Thanks for the clarification John,
I still don't understand the full complexity, It is already possible to
define include paths and macros from within the platform, This should be
enough for a first stage (Many IDE's do that). Then plugins can then write
extentions to change these definition automaticly from the make files or
whatever they like, at least the basics will be there and writting these
plugins shouldn't be too hard as they should be written for each tool.
LE
> The CDT is toolchain agnostic, which means that the parser is expected to
> work for most specified compilers. Even if the compilers are ANSI compliant,
> there can still be problems.
> There is a huge amount of verification we need to do to get things like
> STDIO.H parsing on different platforms, due to the amount of internal
> compiler macro definitions. If you use gcc, for example, and do 'gcc -v',
> you'll see it reads in its specs from some text file in your install that
> tells the compiler what extra #defines to add and re-order the include path.
> Getting this to work for Standard Make projects is very difficult given
> these requirements, as we have no understanding of what toolchain is
> provided. We are working quite diligently to provide solutions in the
> future that will provide code assist/search/refactoring features as good as
> the JDT.
> JohnC
> "Eli" <elil@amdocs.com> wrote in message news:bkpfqf$306$1@eclipse.org...
> > This is a real pity as it means that our company will not be able to use
> > the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse 3.0 it
> > will be more then a year, all this to get a feature that are considered
> > minimum for developers. Is there an alternative?
> >
> > The ast parser is already running, so it shouldn't be too hard to find out
> > the type or am I completely wrong here?
> >
> > Anyway apart from this the CDT has really made some great steps forward.
> >
> > LE
> >
> >
> > > Completing after a "." requires parsing the current document and knowing
> > > the type of the variable before the ".". This is not available now for
> > > Code Assist (release CDT 1.2, October 2003); however, it will be
> available
> > > on the next release (CDT 2.0).
> > > In this release, inside the body of a function or a method, you are able
> > > to find completion proposals of global functions, variables, enums,
> > > macros, etc...
> > > Code assist is using the search feature to find these proposals.
> Anything
> > > that is globally defined that starts with the given prefix will be
> > > searched for and added as a completion proposal. This requires two
> things:
> > > 1- that you have the new indexer enabled ( it will be enabled by default
> > > by the release date, but right now you have to check the "C/C++ Project"
> > > tab for your project's properties. )
> > > 2- that your code is parsing without errors and that your code is saved.
> > > If you have some parsing errors, you do not garantee to find all
> possible
> > > proposals.
> >
> > > symme7ry wrote:
> >
> > > > I've seen previous questions asking about 'code assist' in general,
> but
> > > > not specifically about 'intellisense'.
> >
> > > > When I type "string s; s.", I'd like a bunch of method signatures to
> pop
> > > > up after I type the period. When I type "getline(", I'd like the
> method
> > > > signature to pop up after I type the parenthesis. Does this particular
> > > > feature exist now? Or anything similar? I don't really care that if I
> type
> > > > "f" then control space enter, it fills in a for loop for me. That kind
> of
> > > > code assist I dont care for. I just need some sort of built-in
> knowledge
> > > > of the c++ library. If this does not yet exist, does anyone know when
> it
> > > > will?
> >
> > > > I have eclpise version 2.1.1 and cdt version 1.2.0.51
> >
> >
|
|
|
| Re: 'intellisense' [message #80982 is a reply to message #80938] |
Wed, 24 September 2003 09:03 |
| Eclipse User |
|
|
|
Macro definitions and include path ordering can change depending on the
compiler options in GNU. Our strategy for parsing was to basically
implement the front end of a compiler, with Symbol table and scoping rules
and all of the complexity that comes from the C++ language. This, in the
long term, will yield the correct solution for complex features like
refactoring, writable DOM, search and code assist.
Regarding having extension points edit the configuration for parsing ... we
already have that. CDT is toolchain agnostic and plugins that implement that
extension point need to know specifically what tools are being used, so
that:
a) include paths/macros get set up properly on a compilation unit
granularity
b) the proper parser is discovered (via extension points) to handle that
particular flavour of C/C++, as nearly every compiler introduces its own
constructs; our experience has shown that writing one C++ parser that will
support the union of all of these C++/C variant languages does not work due
to ambiguities and complexities
We plan to set up a target model (the tool representation) post 1.2.
Feel free to send me an email if you are willing to chip in and help.
Thanks
JohnC
"Eli Lopian" <elil@amdocs.com> wrote in message
news:bkrdn9$rpl$1@eclipse.org...
> Thanks for the clarification John,
>
> I still don't understand the full complexity, It is already possible to
> define include paths and macros from within the platform, This should be
> enough for a first stage (Many IDE's do that). Then plugins can then write
> extentions to change these definition automaticly from the make files or
> whatever they like, at least the basics will be there and writting these
> plugins shouldn't be too hard as they should be written for each tool.
>
> LE
>
> > The CDT is toolchain agnostic, which means that the parser is expected
to
> > work for most specified compilers. Even if the compilers are ANSI
compliant,
> > there can still be problems.
>
> > There is a huge amount of verification we need to do to get things like
> > STDIO.H parsing on different platforms, due to the amount of internal
> > compiler macro definitions. If you use gcc, for example, and do
'gcc -v',
> > you'll see it reads in its specs from some text file in your install
that
> > tells the compiler what extra #defines to add and re-order the include
path.
>
> > Getting this to work for Standard Make projects is very difficult given
> > these requirements, as we have no understanding of what toolchain is
> > provided. We are working quite diligently to provide solutions in the
> > future that will provide code assist/search/refactoring features as good
as
> > the JDT.
>
> > JohnC
>
> > "Eli" <elil@amdocs.com> wrote in message
news:bkpfqf$306$1@eclipse.org...
> > > This is a real pity as it means that our company will not be able to
use
> > > the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse 3.0
it
> > > will be more then a year, all this to get a feature that are
considered
> > > minimum for developers. Is there an alternative?
> > >
> > > The ast parser is already running, so it shouldn't be too hard to find
out
> > > the type or am I completely wrong here?
> > >
> > > Anyway apart from this the CDT has really made some great steps
forward.
> > >
> > > LE
> > >
> > >
> > > > Completing after a "." requires parsing the current document and
knowing
> > > > the type of the variable before the ".". This is not available now
for
> > > > Code Assist (release CDT 1.2, October 2003); however, it will be
> > available
> > > > on the next release (CDT 2.0).
> > > > In this release, inside the body of a function or a method, you are
able
> > > > to find completion proposals of global functions, variables, enums,
> > > > macros, etc...
> > > > Code assist is using the search feature to find these proposals.
> > Anything
> > > > that is globally defined that starts with the given prefix will be
> > > > searched for and added as a completion proposal. This requires two
> > things:
> > > > 1- that you have the new indexer enabled ( it will be enabled by
default
> > > > by the release date, but right now you have to check the "C/C++
Project"
> > > > tab for your project's properties. )
> > > > 2- that your code is parsing without errors and that your code is
saved.
> > > > If you have some parsing errors, you do not garantee to find all
> > possible
> > > > proposals.
> > >
> > > > symme7ry wrote:
> > >
> > > > > I've seen previous questions asking about 'code assist' in
general,
> > but
> > > > > not specifically about 'intellisense'.
> > >
> > > > > When I type "string s; s.", I'd like a bunch of method signatures
to
> > pop
> > > > > up after I type the period. When I type "getline(", I'd like the
> > method
> > > > > signature to pop up after I type the parenthesis. Does this
particular
> > > > > feature exist now? Or anything similar? I don't really care that
if I
> > type
> > > > > "f" then control space enter, it fills in a for loop for me. That
kind
> > of
> > > > > code assist I dont care for. I just need some sort of built-in
> > knowledge
> > > > > of the c++ library. If this does not yet exist, does anyone know
when
> > it
> > > > > will?
> > >
> > > > > I have eclpise version 2.1.1 and cdt version 1.2.0.51
> > >
> > >
>
>
|
|
|
| Re: 'intellisense' [message #81127 is a reply to message #80982] |
Thu, 25 September 2003 22:13 |
| Eclipse User |
|
|
|
Hi,all:
Anyway, for a developer using IDE, maybe it's a killer feature when we
type a dot, the methods and attributes of the class before the dot would
list below.
For C++, parsing work may be complex, we hope CDT could add some simple
parsings for user's class members in 1.2 or later, but not very long time
for waiting.
Thanks!
Qin Feng
John Camelon wrote:
> Macro definitions and include path ordering can change depending on the
> compiler options in GNU. Our strategy for parsing was to basically
> implement the front end of a compiler, with Symbol table and scoping rules
> and all of the complexity that comes from the C++ language. This, in the
> long term, will yield the correct solution for complex features like
> refactoring, writable DOM, search and code assist.
> Regarding having extension points edit the configuration for parsing ... we
> already have that. CDT is toolchain agnostic and plugins that implement that
> extension point need to know specifically what tools are being used, so
> that:
> a) include paths/macros get set up properly on a compilation unit
> granularity
> b) the proper parser is discovered (via extension points) to handle that
> particular flavour of C/C++, as nearly every compiler introduces its own
> constructs; our experience has shown that writing one C++ parser that will
> support the union of all of these C++/C variant languages does not work due
> to ambiguities and complexities
> We plan to set up a target model (the tool representation) post 1.2.
> Feel free to send me an email if you are willing to chip in and help.
> Thanks
> JohnC
> "Eli Lopian" <elil@amdocs.com> wrote in message
> news:bkrdn9$rpl$1@eclipse.org...
> > Thanks for the clarification John,
> >
> > I still don't understand the full complexity, It is already possible to
> > define include paths and macros from within the platform, This should be
> > enough for a first stage (Many IDE's do that). Then plugins can then write
> > extentions to change these definition automaticly from the make files or
> > whatever they like, at least the basics will be there and writting these
> > plugins shouldn't be too hard as they should be written for each tool.
> >
> > LE
> >
> > > The CDT is toolchain agnostic, which means that the parser is expected
> to
> > > work for most specified compilers. Even if the compilers are ANSI
> compliant,
> > > there can still be problems.
> >
> > > There is a huge amount of verification we need to do to get things like
> > > STDIO.H parsing on different platforms, due to the amount of internal
> > > compiler macro definitions. If you use gcc, for example, and do
> 'gcc -v',
> > > you'll see it reads in its specs from some text file in your install
> that
> > > tells the compiler what extra #defines to add and re-order the include
> path.
> >
> > > Getting this to work for Standard Make projects is very difficult given
> > > these requirements, as we have no understanding of what toolchain is
> > > provided. We are working quite diligently to provide solutions in the
> > > future that will provide code assist/search/refactoring features as good
> as
> > > the JDT.
> >
> > > JohnC
> >
> > > "Eli" <elil@amdocs.com> wrote in message
> news:bkpfqf$306$1@eclipse.org...
> > > > This is a real pity as it means that our company will not be able to
> use
> > > > the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse 3.0
> it
> > > > will be more then a year, all this to get a feature that are
> considered
> > > > minimum for developers. Is there an alternative?
> > > >
> > > > The ast parser is already running, so it shouldn't be too hard to find
> out
> > > > the type or am I completely wrong here?
> > > >
> > > > Anyway apart from this the CDT has really made some great steps
> forward.
> > > >
> > > > LE
> > > >
> > > >
> > > > > Completing after a "." requires parsing the current document and
> knowing
> > > > > the type of the variable before the ".". This is not available now
> for
> > > > > Code Assist (release CDT 1.2, October 2003); however, it will be
> > > available
> > > > > on the next release (CDT 2.0).
> > > > > In this release, inside the body of a function or a method, you are
> able
> > > > > to find completion proposals of global functions, variables, enums,
> > > > > macros, etc...
> > > > > Code assist is using the search feature to find these proposals.
> > > Anything
> > > > > that is globally defined that starts with the given prefix will be
> > > > > searched for and added as a completion proposal. This requires two
> > > things:
> > > > > 1- that you have the new indexer enabled ( it will be enabled by
> default
> > > > > by the release date, but right now you have to check the "C/C++
> Project"
> > > > > tab for your project's properties. )
> > > > > 2- that your code is parsing without errors and that your code is
> saved.
> > > > > If you have some parsing errors, you do not garantee to find all
> > > possible
> > > > > proposals.
> > > >
> > > > > symme7ry wrote:
> > > >
> > > > > > I've seen previous questions asking about 'code assist' in
> general,
> > > but
> > > > > > not specifically about 'intellisense'.
> > > >
> > > > > > When I type "string s; s.", I'd like a bunch of method signatures
> to
> > > pop
> > > > > > up after I type the period. When I type "getline(", I'd like the
> > > method
> > > > > > signature to pop up after I type the parenthesis. Does this
> particular
> > > > > > feature exist now? Or anything similar? I don't really care that
> if I
> > > type
> > > > > > "f" then control space enter, it fills in a for loop for me. That
> kind
> > > of
> > > > > > code assist I dont care for. I just need some sort of built-in
> > > knowledge
> > > > > > of the c++ library. If this does not yet exist, does anyone know
> when
> > > it
> > > > > > will?
> > > >
> > > > > > I have eclpise version 2.1.1 and cdt version 1.2.0.51
> > > >
> > > >
> >
> >
|
|
|
| Re: 'intellisense' [message #81158 is a reply to message #81127] |
Fri, 26 September 2003 13:09 |
| Eclipse User |
|
|
|
Point taken.
JohnC
"Qin Feng" <qf0421@hotmail.com> wrote in message
news:bl07bo$ghr$1@eclipse.org...
> Hi,all:
> Anyway, for a developer using IDE, maybe it's a killer feature when we
> type a dot, the methods and attributes of the class before the dot would
> list below.
> For C++, parsing work may be complex, we hope CDT could add some simple
> parsings for user's class members in 1.2 or later, but not very long time
> for waiting.
>
> Thanks!
> Qin Feng
>
> John Camelon wrote:
>
> > Macro definitions and include path ordering can change depending on the
> > compiler options in GNU. Our strategy for parsing was to basically
> > implement the front end of a compiler, with Symbol table and scoping
rules
> > and all of the complexity that comes from the C++ language. This, in the
> > long term, will yield the correct solution for complex features like
> > refactoring, writable DOM, search and code assist.
>
> > Regarding having extension points edit the configuration for parsing ...
we
> > already have that. CDT is toolchain agnostic and plugins that implement
that
> > extension point need to know specifically what tools are being used, so
> > that:
>
> > a) include paths/macros get set up properly on a compilation unit
> > granularity
> > b) the proper parser is discovered (via extension points) to handle that
> > particular flavour of C/C++, as nearly every compiler introduces its own
> > constructs; our experience has shown that writing one C++ parser that
will
> > support the union of all of these C++/C variant languages does not work
due
> > to ambiguities and complexities
>
> > We plan to set up a target model (the tool representation) post 1.2.
> > Feel free to send me an email if you are willing to chip in and help.
>
> > Thanks
> > JohnC
>
>
> > "Eli Lopian" <elil@amdocs.com> wrote in message
> > news:bkrdn9$rpl$1@eclipse.org...
> > > Thanks for the clarification John,
> > >
> > > I still don't understand the full complexity, It is already possible
to
> > > define include paths and macros from within the platform, This should
be
> > > enough for a first stage (Many IDE's do that). Then plugins can then
write
> > > extentions to change these definition automaticly from the make files
or
> > > whatever they like, at least the basics will be there and writting
these
> > > plugins shouldn't be too hard as they should be written for each tool.
> > >
> > > LE
> > >
> > > > The CDT is toolchain agnostic, which means that the parser is
expected
> > to
> > > > work for most specified compilers. Even if the compilers are ANSI
> > compliant,
> > > > there can still be problems.
> > >
> > > > There is a huge amount of verification we need to do to get things
like
> > > > STDIO.H parsing on different platforms, due to the amount of
internal
> > > > compiler macro definitions. If you use gcc, for example, and do
> > 'gcc -v',
> > > > you'll see it reads in its specs from some text file in your install
> > that
> > > > tells the compiler what extra #defines to add and re-order the
include
> > path.
> > >
> > > > Getting this to work for Standard Make projects is very difficult
given
> > > > these requirements, as we have no understanding of what toolchain is
> > > > provided. We are working quite diligently to provide solutions in
the
> > > > future that will provide code assist/search/refactoring features as
good
> > as
> > > > the JDT.
> > >
> > > > JohnC
> > >
> > > > "Eli" <elil@amdocs.com> wrote in message
> > news:bkpfqf$306$1@eclipse.org...
> > > > > This is a real pity as it means that our company will not be able
to
> > use
> > > > > the CDT until 2.0 comes out, ans as 2.0 will be written on eclipse
3.0
> > it
> > > > > will be more then a year, all this to get a feature that are
> > considered
> > > > > minimum for developers. Is there an alternative?
> > > > >
> > > > > The ast parser is already running, so it shouldn't be too hard to
find
> > out
> > > > > the type or am I completely wrong here?
> > > > >
> > > > > Anyway apart from this the CDT has really made some great steps
> > forward.
> > > > >
> > > > > LE
> > > > >
> > > > >
> > > > > > Completing after a "." requires parsing the current document and
> > knowing
> > > > > > the type of the variable before the ".". This is not available
now
> > for
> > > > > > Code Assist (release CDT 1.2, October 2003); however, it will be
> > > > available
> > > > > > on the next release (CDT 2.0).
> > > > > > In this release, inside the body of a function or a method, you
are
> > able
> > > > > > to find completion proposals of global functions, variables,
enums,
> > > > > > macros, etc...
> > > > > > Code assist is using the search feature to find these proposals.
> > > > Anything
> > > > > > that is globally defined that starts with the given prefix will
be
> > > > > > searched for and added as a completion proposal. This requires
two
> > > > things:
> > > > > > 1- that you have the new indexer enabled ( it will be enabled by
> > default
> > > > > > by the release date, but right now you have to check the "C/C++
> > Project"
> > > > > > tab for your project's properties. )
> > > > > > 2- that your code is parsing without errors and that your code
is
> > saved.
> > > > > > If you have some parsing errors, you do not garantee to find all
> > > > possible
> > > > > > proposals.
> > > > >
> > > > > > symme7ry wrote:
> > > > >
> > > > > > > I've seen previous questions asking about 'code assist' in
> > general,
> > > > but
> > > > > > > not specifically about 'intellisense'.
> > > > >
> > > > > > > When I type "string s; s.", I'd like a bunch of method
signatures
> > to
> > > > pop
> > > > > > > up after I type the period. When I type "getline(", I'd like
the
> > > > method
> > > > > > > signature to pop up after I type the parenthesis. Does this
> > particular
> > > > > > > feature exist now? Or anything similar? I don't really care
that
> > if I
> > > > type
> > > > > > > "f" then control space enter, it fills in a for loop for me.
That
> > kind
> > > > of
> > > > > > > code assist I dont care for. I just need some sort of built-in
> > > > knowledge
> > > > > > > of the c++ library. If this does not yet exist, does anyone
know
> > when
> > > > it
> > > > > > > will?
> > > > >
> > > > > > > I have eclpise version 2.1.1 and cdt version 1.2.0.51
> > > > >
> > > > >
> > >
> > >
>
>
|
|
|
| Re: 'intellisense' [message #81423 is a reply to message #80982] |
Tue, 30 September 2003 12:40 |
| Eclipse User |
|
|
|
Originally posted by: roca.us.ibm.com
John Camelon wrote:
> Macro definitions and include path ordering can change depending on the
> compiler options in GNU. Our strategy for parsing was to basically
> implement the front end of a compiler, with Symbol table and scoping rules
> and all of the complexity that comes from the C++ language. This, in the
> long term, will yield the correct solution for complex features like
> refactoring, writable DOM, search and code assist.
>
> Regarding having extension points edit the configuration for parsing ... we
> already have that. CDT is toolchain agnostic and plugins that implement that
> extension point need to know specifically what tools are being used, so
> that:
>
> a) include paths/macros get set up properly on a compilation unit
> granularity
> b) the proper parser is discovered (via extension points) to handle that
> particular flavour of C/C++, as nearly every compiler introduces its own
> constructs; our experience has shown that writing one C++ parser that will
> support the union of all of these C++/C variant languages does not work due
> to ambiguities and complexities
You've set yourself a gigantic task! I hope you're aware of GLR parser
generators like Elkhound (http://www.cs.berkeley.edu/~smcpeak/elkhound/)
that make parsing C++ a lot less daunting.
Rob
|
|
|
Goto Forum:
Current Time: Sun Jul 27 05:30:21 EDT 2025
Powered by FUDForum. Page generated in 0.87590 seconds |
Home