CEDET

Darren Hoo writes:
> $ src/emacs -Q -l report-it.el 
>
> then in boolobject.c, try semantic-ia-fast-jump on PyObject
> will enter in to the debugger with a very long backtrace.

Thanks for the detailed recipe, but I'm afraid I can't reproduce it with
Emacs from bzr (revno. 103416) and CEDET from bzr (revno. 8038). If
you're using an older CEDET version, please try the latest development
checkout, which you can get using

bzr checkout bzr://cedet.bzr.sourceforge.net/bzrroot/cedet/code/trunk cedet

Also, delete all cached tags in ~/.semanticdb.

-David

Hi David,

David Engster <deng@...> writes:

> Thanks for the detailed recipe, but I'm afraid I can't reproduce it with
> Emacs from bzr (revno. 103416) and CEDET from bzr (revno. 8038). If
> you're using an older CEDET version, please try the latest development
> checkout, which you can get using
>
> bzr checkout bzr://cedet.bzr.sourceforge.net/bzrroot/cedet/code/trunk cedet

I am using the lastest bzr (rev 8038), I found it also reproducible
with emacs23,  it is ok with cedet bundled with emacs23(seems the
version is (1.0pre7), So I tried  bisecting to track down which
commit exactly causes the problem, I am not quite handy with bzr's
bisect, So I even checkouted git mirror of cedet on github, but when 
I try to build some(about 5 )previous revisions, No one can be 
successfully built, for example with revno: 7949 

Author: zappo <zappo@...>
Date:   Sat Jul 31 01:17:35 2010 +0000

    (pulse-flag): Update doc.
    (pulse-momentary-highlight-overlay): If pulse-flag is 'never, then
    disable all pulsing.

Got some errors like this:

make[2]: Leaving directory `/tmp/cedet/semantic/ctags'
make -C bovine
make[2]: Entering directory `/tmp/cedet/semantic/bovine'
"emacs" -batch --no-site-file -l grammar-make-script -f semantic-grammar-batch-build-packages c.by make.by scheme.by erlang.by
Compiling Grammars from: /tmp/cedet/semantic/semantic-grammar.elc
Wrong type argument: listp, $1
Args out of range: ",", 1
Wrong type argument: listp, $2
Wrong type argument: listp, $2
make[2]: *** [by] Error 1
make[2]: Leaving directory `/tmp/cedet/semantic/bovine'
make[1]: *** [bovinator] Error 2
make[1]: Leaving directory `/tmp/cedet/semantic'
make: *** [semantic] Error 2

So for the time being I got the the following test result:

Emacs         cedet                 result
---------------------------------------------------
23.2        bundled (1.0pre7)       OK
23.2        bzr 8038                fail|reproducible
trunk       bundled (1.0)           fail|reproducible
trunk       bzr 8038                fail|reproducible
trunk       several old revision    fail to build


> Also, delete all cached tags in ~/.semanticdb.

Yes, I never save cached tags in my test enviroment.

> Wrong type argument: listp, $1
> Args out of range: ",", 1
> Wrong type argument: listp, $2
> Wrong type argument: listp, $2
> make[2]: *** [by] Error 1
>
After going through the mailinglist, it's good to know that 
this is caused by a change in lread.c. Now I can track down
my bug using emacs23 and pinpoint the commit that introduced
it:

$ git bisect bad
e455de8cb19834cc142e0ef6042a35faa131b520 is the first bad commit
commit e455de8cb19834cc142e0ef6042a35faa131b520
Author: zappo <zappo@...>
Date:   Fri Feb 19 02:23:27 2010 +0000

    (semantic-expand-c-tag-namelist): Split a full type from a typedef out
    to its own tag.

I've create a simple case in case Python source is 
not at your hand. 

save the attachments to the directory /tmp

and change first line in t.el to point to your
cedet bzr build. the sample code t.c is extracted 
from the python source code simplified.

invoke emacs as 

$emacs23 -Q -l /tmp/t.el

then M-x  test-it should show a max-specpdl-size error.

Darren Hoo writes:
> $ git bisect bad
> e455de8cb19834cc142e0ef6042a35faa131b520 is the first bad commit
> commit e455de8cb19834cc142e0ef6042a35faa131b520
> Author: zappo <zappo@...>
> Date:   Fri Feb 19 02:23:27 2010 +0000
>
>     (semantic-expand-c-tag-namelist): Split a full type from a typedef out
>     to its own tag.

Hmm. Even with your reduced recipe I cannot reproduce that
problem. Could you please send the full backtrace directly to my mail
address?

-David

Hi David,

Your last commit (8040) has fixed this problem. I just want
to confirm it. Well done and thanks a lot.

cheers.

Darren Hoo writes:
> Your last commit (8040) has fixed this problem. I just want
> to confirm it. Well done and thanks a lot.

Great. Thanks for letting me know and for providing all those
recipes. This was rather hard to reproduce (and fix...).

-David