CEDET

Hi Michael,

Thanks for following up on your question.  I'm not familiar with the 
problem you related.  If you have calls to semantic-add-system-include 
for c-mode similar to those for c++ mode, then things should work ok. 
There is some trickyness in how major modes inherit from each other that 
makes c-mode and c++-mode potentially confusing.  Usually when c-mode 
features appear in c++-mode, or not.  It may also be that when searching 
for includes, it didn't want to pick up C files for use with C++.  That 
is supposed to work, so if you have an example, that would be 
interesting to see.

Eric

Vance, Michael wrote:
> Hm, managed to resolve this by appending my project dirs onto the 
> auto-mode-alist with c++-mode so that headers would open in c++-mode 
> rather than c-mode; not sure why adding the system includes for c-mode 
> also didn’t have the same effect, I’d assume they would.
> 
>  
> 
> Thanks,
> 
>  
> 
> m.
> 
>  
> 
> *From:* Vance, Michael
> *Sent:* Friday, December 11, 2009 2:48 PM
> *To:* 'cedet-semantic@...'
> *Subject:* system include paths
> 
>  
> 
> Hi all,
> 
>  
> 
> I have a quick question regarding system include paths. I have an 
> ~/.emacs file with a variety of lines of the form:
> 
>  
> 
> ;; CEDET / Semantic / Xbox 360 system includes
> 
> (semantic-add-system-include "c:/Program Files (x86)/Microsoft Xbox 360 
> SDK/include/xbox" 'c++-mode)
> 
>  
> 
> If I open a file outside a project root, it will find headers in the 
> system include dirs just fine. However if I open up a file inside an 
> ede-cpp-root-project, I’ll get tons of unparsed includes. When I look at 
> their summary page I’ll see something like below. Is the 
> ede-cpp-root-project somehow masking the Semantic system include? Do I 
> need to somehow pass my Semantic system include list to the 
> ede-cpp-project-root as :system-include or some such? I tried doing this 
> with semantic-dependency-system-include-path but it didn’t seem to work 
> properly. I’ve also tried duplicating all my semantic-add-system-include 
> calls for c-mode as well as c++-mode but that didn’t seem to work either.
> 
>  
> 
> Thanks,
> 
>  
> 
> m.
> 
>  
> 
> Include Summary for File: d:/jobqueue/xenon/jobqueue_xenon.h
> 
>  
> 
> This file contains 17 tags, 3 of which are includes.
> 
>    Unknown Includes:  2
> 
>    Unparsed Includes: 0
> 
>    Parsed Includes:   1
> 
>  
> 
> Include Path Summary:
> 
>  
> 
>   This file's project include search is handled by the EDE object:
> 
>     Buffer Target:  #<ede-cpp-root-target d:/jobqueue/xenon/>
> 
>     Buffer Project: #<ede-cpp-root-project jobqueue>
> 
>     Backup in-project Locator: #<ede-locate-base Loc>
> 
>     EDE Project system include path: Empty
> 
>  
> 
>   This file's system include path is:
> 
>     /usr/include
> 
>  
> 
> All unknown includes:
> 
>   xmmintrin.h
> 
>   types.h
> 
>   stdint.h
> 
>   time.h
> 
>   stdlib.h
> 
>   spu_printf.h
> 
>   spu_intrinsics.h
> 
>   spu_mfcio.h
> 
>   sys/time_util.h
> 
>   ppu_intrinsics.h
> 
>   ppcintrinsics.h
> 
>   xtl.h
> 
>   windows.h
> 
>  
> 
> There are currently no includes found recursively.
> 
>  
> 
> [back]
> 
> 
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________
> 
> 
> ------------------------------------------------------------------------
> 
> ------------------------------------------------------------------------------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> cedet-semantic mailing list
> cedet-semantic@...
> https://lists.sourceforge.net/lists/listinfo/cedet-semantic

My example is fairly vanilla, although I'm not sure if there's a way to
cleanly package it so I'll just show the forensics.

An excerpt from my .emacs:

(semantic-add-system-include "c:/usr/local/cell/target/spu/include"
'c++-mode)
(semantic-add-system-include
"c:/usr/local/cell/target/spu/include/cell/spurs" 'c++-mode)
(semantic-add-system-include "c:/usr/local/cell/target/spu/include"
'c-mode)
(semantic-add-system-include
"c:/usr/local/cell/target/spu/include/cell/spurs" 'c-mode)

;; CEDET / EDE / Foo
(add-to-list 'auto-mode-alist (cons "d:/foo" 'c++-mode))
(ede-cpp-root-project "foo"
		      :name "foo"
		      :file "d:/foo/main/Jamfile"
		      :include-path (split-string
(shell-command-to-string "dir /b /s /ad
\"d:\\foo\\main\\install\\ps3-gcc-dbg\"") "\n" t)
)

I then open a file, let's say foo.cpp, under d:\foo that contains an
include in the first system-include dir above (spurs.h). From foo.cpp I
can successfully visit the header (ie, it is parsed). However from that
header I cannot visit anything *it* includes from that hierarchy. If I
do a summarize I see:

Include Summary for File:
c:/usr/local/cell/target/common/include/cell/spurs.h

This file contains 29 tags, 28 of which are includes.
   Unknown Includes:  28
   Unparsed Includes: 0
   Parsed Includes:   0

Include Path Summary:


  This file's system include path is:
    /usr/include

There are currently no includes found recursively.

If I then switch that buffer (spurs.h) to c++-mode it automatically
switches/re-parses/whatever everything correctly and now when I do a
summarize I see what I expect:

Include Summary for File:
c:/usr/local/cell/target/common/include/cell/spurs.h

External Database References to this buffer:
    jobqueue.cpp
    jobqueue_ps3.h
This file contains 29 tags, 28 of which are includes.
   Unknown Includes:  0
   Unparsed Includes: 0
   Parsed Includes:   28

Include Path Summary:


  This file's system include path is:
[...]
    c:/usr/local/cell/target/spu/include/
    c:/usr/local/cell/target/spu/include/cell/spurs/
[...]

I'll think about building a little test version of this when I have some
time later this week, but if you have any clever diagnostics for me to
try just let me know.

Thanks,

m.

-----Original Message-----
From: Eric M. Ludlam [mailto:eric@...] 
Sent: Sunday, December 13, 2009 8:53 PM
To: Vance, Michael
Cc: cedet-semantic@...
Subject: Re: [cedet-semantic] system include paths

Hi Michael,

Thanks for following up on your question.  I'm not familiar with the 
problem you related.  If you have calls to semantic-add-system-include 
for c-mode similar to those for c++ mode, then things should work ok. 
There is some trickyness in how major modes inherit from each other that

makes c-mode and c++-mode potentially confusing.  Usually when c-mode 
features appear in c++-mode, or not.  It may also be that when searching

for includes, it didn't want to pick up C files for use with C++.  That 
is supposed to work, so if you have an example, that would be 
interesting to see.

Eric

Vance, Michael wrote:
> Hm, managed to resolve this by appending my project dirs onto the 
> auto-mode-alist with c++-mode so that headers would open in c++-mode 
> rather than c-mode; not sure why adding the system includes for c-mode

> also didn't have the same effect, I'd assume they would.
> 
>  
> 
> Thanks,
> 
>  
> 
> m.
> 
>  
> 
> *From:* Vance, Michael
> *Sent:* Friday, December 11, 2009 2:48 PM
> *To:* 'cedet-semantic@...'
> *Subject:* system include paths
> 
>  
> 
> Hi all,
> 
>  
> 
> I have a quick question regarding system include paths. I have an 
> ~/.emacs file with a variety of lines of the form:
> 
>  
> 
> ;; CEDET / Semantic / Xbox 360 system includes
> 
> (semantic-add-system-include "c:/Program Files (x86)/Microsoft Xbox
360 
> SDK/include/xbox" 'c++-mode)
> 
>  
> 
> If I open a file outside a project root, it will find headers in the 
> system include dirs just fine. However if I open up a file inside an 
> ede-cpp-root-project, I'll get tons of unparsed includes. When I look
at 
> their summary page I'll see something like below. Is the 
> ede-cpp-root-project somehow masking the Semantic system include? Do I

> need to somehow pass my Semantic system include list to the 
> ede-cpp-project-root as :system-include or some such? I tried doing
this 
> with semantic-dependency-system-include-path but it didn't seem to
work 
> properly. I've also tried duplicating all my
semantic-add-system-include 
> calls for c-mode as well as c++-mode but that didn't seem to work
either.
> 
>  
> 
> Thanks,
> 
>  
> 
> m.
> 
>  
> 
> Include Summary for File: d:/jobqueue/xenon/jobqueue_xenon.h
> 
>  
> 
> This file contains 17 tags, 3 of which are includes.
> 
>    Unknown Includes:  2
> 
>    Unparsed Includes: 0
> 
>    Parsed Includes:   1
> 
>  
> 
> Include Path Summary:
> 
>  
> 
>   This file's project include search is handled by the EDE object:
> 
>     Buffer Target:  #<ede-cpp-root-target d:/jobqueue/xenon/>
> 
>     Buffer Project: #<ede-cpp-root-project jobqueue>
> 
>     Backup in-project Locator: #<ede-locate-base Loc>
> 
>     EDE Project system include path: Empty
> 
>  
> 
>   This file's system include path is:
> 
>     /usr/include
> 
>  
> 
> All unknown includes:
> 
>   xmmintrin.h
> 
>   types.h
> 
>   stdint.h
> 
>   time.h
> 
>   stdlib.h
> 
>   spu_printf.h
> 
>   spu_intrinsics.h
> 
>   spu_mfcio.h
> 
>   sys/time_util.h
> 
>   ppu_intrinsics.h
> 
>   ppcintrinsics.h
> 
>   xtl.h
> 
>   windows.h
> 
>  
> 
> There are currently no includes found recursively.
> 
>  
> 
> [back]
> 
> 
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________
> 
> 
>
------------------------------------------------------------------------
> 
>
------------------------------------------------------------------------
------
> Return on Information:
> Google Enterprise Search pays you back
> Get the facts.
> http://p.sf.net/sfu/google-dev2dev
> 
> 
>
------------------------------------------------------------------------
> 
> _______________________________________________
> cedet-semantic mailing list
> cedet-semantic@...
> https://lists.sourceforge.net/lists/listinfo/cedet-semantic

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________

Vance, Michael wrote:
> My example is fairly vanilla, although I'm not sure if there's a way to
> cleanly package it so I'll just show the forensics.
> 
> An excerpt from my .emacs:
> 
> (semantic-add-system-include "c:/usr/local/cell/target/spu/include"
> 'c++-mode)
> (semantic-add-system-include
> "c:/usr/local/cell/target/spu/include/cell/spurs" 'c++-mode)
> (semantic-add-system-include "c:/usr/local/cell/target/spu/include"
> 'c-mode)
> (semantic-add-system-include
> "c:/usr/local/cell/target/spu/include/cell/spurs" 'c-mode)
> 
> ;; CEDET / EDE / Foo
> (add-to-list 'auto-mode-alist (cons "d:/foo" 'c++-mode))
> (ede-cpp-root-project "foo"
> 		      :name "foo"
> 		      :file "d:/foo/main/Jamfile"
> 		      :include-path (split-string
> (shell-command-to-string "dir /b /s /ad
> \"d:\\foo\\main\\install\\ps3-gcc-dbg\"") "\n" t)
> )

Hi,

I suspect the problem is that the :file slot is pointing to 
d:/foo/main/Jamfile, but that the root of the project is at d:/foo.  It 
should be d:/foo/SOMEFILE, where SOMEFILE could be the directory 'main' 
if needed.  This file is what marks the root of the project.

As such, when you find something in the include path in 
foo/main/install, it isn't part of that project, and the includes cannot 
be chased.

Eric