Re: [brlcad-devel] mged fails regesssion on trunk

[Clifford Y. <cli...@gm...>]

On Sun, Apr 8, 2012 at 12:26 PM, Tom Browder <tom...@gm...> wrote:


> For one thing, I just noticed that "make" env | grepp with autotools
> removes the regress/CMakeLists.txt, that is confusing the issue!
>

Ow.  Yeah, that won't help.


> Okay, I just found a disconnect with the tcl library version--cured
> that with an install BEFORE the regress check (did tcl bump a patch
> number recently?).
>

Wait - which one did you install before running regress?  The CMake
regression in particular shouldn't care in the least whether you installed
or not.


> Now regression failure has changed to this for both builds:
>
> ERROR: relos does not exist!
> Output: invalid command name "relos"
>

Weird.  I'm not seeing that here - will have to figure out what's calling
"relos".


> But, for cmake, the regression fails at that point.
>

OK, that's expected if there is indeed a problem.


> For autotools, there is another regression failure:
>
> /bin/sh ../regress/flawfinder.sh ..
> running flawfinder...
>

Ah, right.  The flawfinder.sh script had been busted, and thus was silently
"passing" by not doing anything.  I ended up "fixing" the script and thus
exposing that we aren't yet set up to use flawfinder.  The correct thing to
do right now is have the autotools regression just not call the flawfinder
regression as part of its regression build target.

Cheers,
CY