On Wed, Oct 21, 2009 at 09:49:49PM +0200, Martin Quinson wrote:
> Le mercredi 21 octobre 2009 à 18:45 +0200, Philipp Niemann a écrit :
> > On Wed, Oct 21, 2009 at 10:00:20AM +0200, Martin Quinson wrote:
> > > scons: *** [build/native-debug/sound/fxset] Source
> > > `src/trigger/trigger.cc' not found, needed by target
> > > `build/native-debug/sound/fxset'.
> > > scons: building terminated because of errors.
> > >
> > > Am I the only one having this message? Did I forget to do something
> > > first?
> > Yep, it seems you're the only one so far.
> I guess my problem comes from this message spit by scons:
> WARNING: utils/spurious_source_code/whitespace_checker does not exist!
> WARNING: Go to utils/spurious_source_code and build it with make!
> WARNING: Else many checks will be done much slower or not at all!
> So I guess I'm falling back to a rather untested method, explaining why
> it failed for me and only for me. So, attached is a patch to the scons
> files mentionning this source file killed back in january 2008. Should I
> send it to the sourceforge tracker or sending it to this list is enough?
Hm, well... I am rather sure the above messages (WARNING: ...) are
completly unrelated to the build error. The messages concern whitespace
checker stuff that is done before any gcc compiles are called (except the
scons-config tests which are done first after starting scons).
> Then, I assume that it's ok when the test cases do fail, right?
Testcases? I don't think I get the point.
There actually are all three things we discuss up to here, all called by
scons at some stage of a build process:
1) whitespace checks
2) compiling the source with c++ compiler
3) testing the created binary with tests (somehow, no idea)
To be precise there are whitespace checks at stage 2) as well, IIRC,
using a different implementation. Might well be stage 1) checks are
deactivated nowadays, I am not sure about this. But your WARNING: messages
above came/come from stage 1). They reference a ADA
re-implementation/upgrade of python scripts that were the first used in
the first stages of whitespace-checking. As they where mostly not-scons
aware the stage 2) white-space checks exist.
Stage 1) and Stage 3) are rather new. I wildly guess: 1) about six-nine
months old, 3) about four-seven month. Whitespace-checks with stage 2)
where introduced after both 1) and 3). All them changed until I lost close
contact to source/the list, which might be about six months ago (long, so
add another six-month security range to all times I state). All in all be
careful about my words. But don't think I stop writing now... ;)
Well, thinking again I will stop here. Any word about your patch is even
wilder guessing. I don't get the point of the patch. And I'm quite sure
you didn't wrote to learn about the history of whitespace checking
I'm afraid I'm of no help or worse. I put it as question: Which stage of
the build process does your error occur in? I just compiled widelands
again and have no idea what could have gone wrong with your source except
falling back to svn re-checkout the complete sourcetree, but Peter already
Have you tried scons distclean? I had to rm -Rf build/ and check that out
again 'cause of some weird scons error after doing distclean but still
this is the closest thing to re-checkout I know of.