Re: [Flex-help] flex-2.5.37 Tests FAILED: 48A

[Dennis C. <dc...@bl...>]

> Hugh,
> 
> It would be hard to maintain without some behind-the-scenes magic that
> itself would be hard to maintain.

Solaris is getting to be more and more of a drag to support also. The nice folks at Oracle have no love for open source and I was one of the guys on the last OpenSolaris governance board when Oracle acquired Sun and then took a large dump all over open source Solaris ( OpenSolaris ). Lucky for us a large number of guys took off out of there all screaming "fork!" at the top of their lungs and we have the Illumos project. 

Does anyone remember these ugly days ?  : 

http://www.serverwatch.com/trends/article.php/3877526/The-Looming-Demise-of-OpenSolaris.htm

If you were not into Solaris .. you wouldn't. But for a brief shining moment we had a fully open production grade commercial UNIX with funding from Sun Microsystems. Then Larry walked in from his yacht ( http://3blondesandaboat.blogspot.ca/2012/08/looking-at-boats.html ) and killed it with a snap of his fingers. 

There are still a lot of good people around however ... still working away, quietly in dark caves while our beards all grow a little longer :-)
 
> Flex documents its test suite dependencies, for what that's worth. It's
> also possible to build flex without having either flex or bison directly
> available. (Well, you need a C compiler, but that's an indirect
> dependency.)

yep .. it is the usual game, no big surprises unless you run "gmake check" and then ka-whammo all the tests fail and most people would stop there and say "no bloody way". 

I have a tendency to clicmb in and .. well, fix it while grumping a bit. Hey, I'm old, been doing this UNIX thing 25 years now so hopefully people cut me some slack :-)

Just about all the Makefiles have this in it now : 

# diff tests/test-multiple-scanners-nr/Makefile.new tests/test-multiple-scanners-nr/Makefile    
391c391
<       $(CC) $(CXXFLAGS) -o $@ $(LDFLAGS) $(OBJS) $(LOADLIBES)
---
>       $(CC) -o $@ $(LDFLAGS) $(OBJS) $(LOADLIBES)

...and yes, that is wrong !  arg ... should be just $(CFLAGS) there. Not $(CXXFLAGS) despite the fact that the change works but only because :

# echo $CXXFLAGS 
-xildoff -m64 -xmemalign=8s -xcode=pic32 -xregs=no%appl -xlibmieee -mc -g -xs -xunroll=1 -xtarget=sparc -xcache=generic -xchip=ultra -D_TS_ERRNO -D_POSIX_PTHREAD_SEMANTICS -D_LARGEFILE64_SOURCE

which works fine on a generic Sparc server. 

So let me get in there and fix my Makefiles and then yes, I will submit a patch but really the patch is just adding in $(CFLAGS) on all the compile lines in those test directories Makefile(s) and then, the tests work as expected. A side note is that maybe it would be nice if the bison tests fail with a message that says "you need bision which needs flex. This is a circular dependency" but that would be spoon feeding, and who wants to do that ?  ;-)

Dennis