Re: [lmi] Fwd: How did this unit test fail?

[Greg Chicares]

On 2018-03-22 15:18, Vadim Zeitlin wrote:
> On Wed, 21 Mar 2018 13:31:14 +0000 Greg Chicares <address@hidden> wrote:
> 
> GC> I'm wondering the problem could be that debian's MinGW-w64 package
                   ^ if
> GC> is missing some upstream fix. I'm using the latest:
> GC> 
> GC> # apt-get -s install g++-mingw-w64
> GC> Reading package lists... Done
> GC> Building dependency tree       
> GC> Reading state information... Done
> GC> g++-mingw-w64 is already the newest version (7.2.0-20+20.2).
> 
>  I have some interesting news: using exactly the same compiler in my Buster
> chroot, I can indeed reproduce the problem, i.e. I see exactly the same
> failure as you do. However for me it happens with the unmodified master
> (a86806ca422d6a40cc5f0424932367ee3aa74cd5) sources, i.e. without reverting
> the changes of 3d3e2b67916c72c7446f140f2f986456b1164b1e. Moreover, if I do
> revert them, then the problem goes away. I.e. I see exactly the converse of
> what you see which is really, really weird.

Fascinating.

>  Part of the explanation is that we use different build systems and so
> (slightly) different compiler flags

Both should use essentially the same flags already. If they don't,
then presumably this debugging exercise will show where they differ,
and then we can make them the same.

OTOH, if both build systems specify the same flags, but in a different
order, then it might be too much work to force identical ordering.
(Of course, '-O0 -O3' and '-O3 -O0' mean different things, but my
point here is that '-W -Wall' should be equivalent to '-Wall -W'
and it shouldn't be necessary to write them in the same order.)

>  As an aside, I am still not sure what do I need to do use the official
> makefiles under Linux because if I just do "make bourn_cast_test.exe", I
> get plenty of errors, e.g. about missing g++ etc. Is there a script similar
> to install_msw.sh for Linux that I could follow, by chance?

Not yet. It'd be really nice to have, but it'll take a lot of work
and I haven't done it yet.

For cross compiling (as opposed to building natively for GNU/Linux),
there's 'README.schroot', which I'm careful to update every time I
use it. It shows exactly what I did to create my "buster" chroot.
It's a long list of copy-pastable snippets rather than an executable
script. And it probably wouldn't help you because you're the person
who showed me how to do the hard parts.

>  Right now I don't understand at all what's going on here, I'm going to try
> to minimize the test case and open a gcc bug report if I manage to do it
> because it clearly looks like something is very, very wrong with it.

Great, thanks, I'll look forward to hearing what you find.