Thanks for the detailed notes. I haven't gotten to them all yet. More
Chong, some of the items may be in the Emacs variant of the doc only.
On 07/26/2010 10:32 PM, Christoph wrote:
> Hi Eric,
> Here is the list of issues found with the current CEDET version in Emacs
> 24.0.50 (trunk):
> - doc-string for 'pulse-flag': both options mentioned refer to non-nil.
> nil option is not mentioned in doc-string.
I fixed this and will check it into CEDET/CVS soon.
> - setting pulse-flag to nil causes cedet to highlight the symbol using
> semantic-ia-fast-jump. However, highlighting does not disappear after a
> short while (as stated in pulse-flag documentation), but stays until the
> next Emacs command.
Right. That is the case where pulsing is not available. I'll fix the
> - In (semantic) 'Semantic mode user commands': `\C-c , up', remove extra
> - (semantic) Semantic Internals: Semanticdb Internals entry in menu has
> additional whitespace, correct alignment with other entries
I'm guessing these are in the Emacs version of the manual as I didn't
find the problems locally.
> - Command 'M-x semantic-mode, then M-x
> global-semantic-idle-completions-mode'. On putting point on symbol
> mini-buffer shows 'Bug Showing Completions: (wrong-type-argument
> number-or-marker-p (+ -4))'
> - In (semantic) SymRef: node refers to variable semantic-symref-tool but
> that variable does not exist.
This variable is in semantic-symref.el, or it is in the CEDET/CVS
version. Perhaps it wasn't loaded?
> Other notes, questions etc.:
> - Can highlighting on semantic-ia-fast-jump be disabled entirely?
It cannot. It could certainly be made into an option though. Are you
looking for this on a global scope, or for individual commands?
> - No indication that Semantic mode or any of its minor modes is active
> in mode line.
Correct. It used to, and the mode line got over full of bits that were
effectively useless. A global 'semantic mode' may be useful though.
The version of CEDET I've been working on doesn't have a semantic mode.
I can usually tell because the decoration mode turns on and it becomes
> - How are "tags" (source-code units deemed "important") defined? How
> does semantic identify these "tags"? How can the user tell what has
> been identified as a "tag"? E.g. `C-c , C-w' on main() kills my
> *entire* main() function.
If senator is on, you can use C-c , ? to blink the tag
> - 'semantic-ia-show-summary': if no symbol information is available
> (e.g. file containing symbol has not been analyzed), function does
> nothing and there is no feedback. Maybe it should report 'No symbol
> information available' in the mini-buffer.
Good idea. I'll check something in for that.
> - Using 'semantic-ia-show-summary' on a #define in C results in the
> following output in the mini-buffer: 'Variables: const MY_DEFINE'
> This is confusing, since the macro is not a variable.
Indeed. The macros are treated as variables for all other purposes.
Since then the architecture has evolved, and I suspect this could be
changed. I'll look into it.
> - 'semantic-analyze-current-context' on static, module-scope function
> (e.g. static void myFunc(int param): Buffer shows the following error
> "Encountered Errors: '(error "Cannot find definition for \"static\"")
This function is good for debugging, so it usually throws an error when
things go wrong. Other tools capture those errors and do something a
little more graceful.
In this case, however, I can't replicate the issue in CEDET/CVS. I
wonder if this is something I had fixed there. Can you provide a more
complete example function and cursor location?