Re: [lmi] An lmi anomaly with gcc-6.3

[Greg Chicares]

On 2017-08-21 11:48, Vadim Zeitlin wrote:
> On Mon, 21 Aug 2017 02:24:17 +0000 Greg Chicares <address@hidden> wrote:
> 
> GC> On 2017-08-21 00:35, Greg Chicares wrote:
> GC> [...]
> GC> > I hope we can abandon gcc-4.9 and just use 6.3 for lmi: 6.3 seems to
> GC> > work perfectly well in a cygwin VM here
> GC> 
> GC> I did see one imperfection when I ran this command:
> GC> 
> GC> logfile=log-`date -u +'%Y%m%dT%H%MZ'`; echo "Log file is 
> '$logfile.tar.bz2'."; ./install_msw.sh >$logfile 2>&1; tar -cjf 
> $logfile.tar.bz2 $logfile
> GC> 
> GC> The log file says:
> [...when making wx_test.exe...]
> GC> cregex.o: In function `...unpronounceable monstrosity...': undefined
> GC> reference to `...another slightly less unpronounceable one...`
> GC> 
> GC> ...followed by four other 'cregex.o' errors, and 'wx_test.exe'
> GC> fails to link. All other programs link correctly. Building
> GC> 'wx_test.exe' later also works, so this is not a critical
> GC> problem.
> 
>  I can't make any sense of it and I can't reproduce this by doing "make
> clean" followed by "make install" neither. I didn't run the full
> install_msw.sh because it takes a rather long time, but I could it, of
> course, it's just that I don't see how could it explain the problem and so
> even if it did allow me to reproduce the problem, I'm not sure it would
> help me to understand it.

I thought it might be correlated with the minimal path used in
'install_msw.sh':

export         PATH=/opt/lmi/local/bin:/opt/lmi/local/lib:$PATH
export 
minimal_path=/opt/lmi/local/bin:/opt/lmi/local/lib:/usr/bin:/bin:/usr/sbin:/sbin
...
make $coefficiency --output-sync=recurse PATH=$minimal_path install

I say "correlated" rather than "caused" because I don't see how
changing $PATH could cause the reported error.

>  I know it's a cop out, but could it have been just a random fluke (not
> sure if cosmic rays are more or less likely on the eclipse days)?

Less, according to
  "Variation of Cosmic Ray Intensity During the Solar Eclipse August 11, 1999"
in Astronomical Society of the Pacific, ASP Conference Series vol. 205.:
  "the mean drop in low energy cosmic ray flux of gamma rays
  has been detected as 11%"

But compare
  "Measurements of Cosmic Rays during the Solar Eclipse of June 19 1936"
in Japanese Journal of Astronomy and Geophysics, Vol. 14, p.265:
  "neither variation in intensity, nor fluctuation of cosmic rays ... occurred"

However, so-called science doesn't provide all the answers, so we
must not rush to dismiss alternative viewpoints that predict...
  
http://www.express.co.uk/news/weird/843095/Eclipse-2017-total-solar-eclipse-warning-Glastonbury-universe
  "great shifts in the timelines of universal consciousness, heralding
  in the emergence and anchoring in of the Golden Age frequencies into
  our planetary grid systems"
but the Glastonbury I live in doesn't have seven hills to dance upon.

At any rate, my ECC RAM should notify me of any harmful cosmic-ray
event while, I hope, letting the Golden Age vibrations through.

> Or is it
> reproducible in your case

Yes, three times out of three.

> and, if so, is it enough to run "make clean" to
> see it happen again or not?

Yes, 'make clean' has detoxified the harmful emanations and brought
about a reharmonization of the build. Or, if you want a "scientific"
explanation, no other 'make' target even used the boost regex stuff,
and re-extracting it from old archives didn't change any of its
source files' dates, so it didn't get rebuilt, and objects built
with gcc-4.9.1 use a different API.