[Scons-dev] scons terminating silently

[Tom Tanner (BLOOMBERG/ LONDON)]

I think I know have a handle on this.

My SConscript is calling env.Subst before the main build process starts. This doesn't offhand seem to be a particularly wrong thing to do. I think it's not unrelated to another problem I have, as our C++ include path and defiens are different to our C ones, and our C++ objects are called with
   env.StaticObject(blah, CPPDEFINES = env['CXXDEFINES'], CPPPATH = env['CXXPATH'] + env['CPATH'])
whereas our C ones are called thus:
   env.StaticObject(blah, CPPDEFINES = env['CDEFINES'], CPPPATH = env['CPATH'])

and if I change Subst.py at line 444 to just raise the exception, with my previously mentioned broken import:

def _concat_dict(blah):
    from SCons.Defaults import _concay

env['_concat'] = _contact_dict

Then the stack trace goes back via bits of Subst to a env.StaticObject call.

So, basically, Subst 'does things' to any exceptions raised during processing, and this calls scons to fail silently. Please could this be fixed. I'm not knowledgable enough to determine whether or not the change I did is at all safe.

env.Subst can under the right circumstances throw a BuildError. I think it's related to the code at lines 444 and / or 666 which translates any raised exception into a BuildError.

As this is happening when reading the SConscript file, this gets caught in Main.py, which promptly prints nothing at all and exits quietly.





----- Original Message -----
From: 
To: garyo at oberbrunner.com, scons-dev at scons.org
At: Aug 24 2012 18:01:07

OK, for some reason, SCons is getting this ImportError thrown whilst processing my SConstruct file. This is throwing a BuildError which has wrapped an ImportError. The code at line 986 only traps StopErrors, so it goes back up to to main code and doesn't print anything (see lines 1380 or so)

----- Original Message -----
From: garyo at oberbrunner.com
To: TOM TANNER (BLOOMBERG/ LONDON)
At: Aug 23 2012 20:56:05

On Tue, Aug 21, 2012 at 3:55 AM, TOM TANNER (BLOOMBERG/ LONDON)
<ttanner2 at bloomberg.net> wrote:

> That helps not at all, sadly


Please do let us know if you figure it out.  A bunch of over-broad
exception handlers were cleaned up a couple of years ago, maybe some
are still lurking.

-- 
Gary

_______________________________________________
Scons-dev mailing list
Scons-dev at scons.org
http://two.pairlist.net/mailman/listinfo/scons-dev