Re: [lmi] [PATCH] wx_test_validate_output.cpp

[Greg Chicares]

On 2015-03-12 00:45, Vadim Zeitlin wrote:
[...]
>  Granted, I still don't know how to fix it, but I can experiment. The
> trouble is that while I can't reproduce it locally, I have to ask you to do
> the tests for everything I try, so I'd like to think a bit more about what
> exactly could be worth trying to avoid wasting your time unnecessarily.

I'll do whatever it takes to work through this with you. I need this fixed.

>  But for now I'm afraid I still want to ask you to do something else, just
> to remove the meagre but not completely impossible possibility that
> something is wrong with your build: could you please also try running the
> executable (built by me inside my lmi VM, so very unlikely to be infected
> or otherwise compromised) from here:
> 
>       https://www.dropbox.com/s/cqmto1o879z2in7/uiaction.tar.bz2
> 
>  Does it also misbehave in the same way?

No: it does not misbehave. Output equals input, as expected:
  pound, one, zero, underbar
It works if I double-click it in "explorer", or if I run it in cygwin zsh.
Then I run the version I built in the same zsh session, and that fails.

Mine should use the wx version given here:
  http://lists.nongnu.org/archive/html/lmi/2015-02/msg00022.html
i.e. 2f239930eb3bc8a928a771d4659e46900907b11f but does it really?
'git log -1 | head -n1' gives 3c0b17d... and I hope that's enough because
it's kind of tedious to write email in the host when I can't copy and
paste from the guest. Looks wrong...so I try:
  cd /lmi/src/lmi
  md5sum /cache_for_lmi/downloads/wxWidgets-3.1.0.tar.bz2
and get 3c049ed... which also doesn't match my expectation. Let's see...
that msg00022 is from 28 February, so let's try:
  ls -o /cache_for_lmi/downloads/wxWidgets-3.1.0.tar.bz2
and it says 30 January. Aha:
  http://lists.nongnu.org/archive/html/lmi/2015-01/msg00043.html
| Update--please set your secret decoder ring to:
|   3c0b17d4b5cbb6c1ac8282f6a1489b6260e09bf3
    ^^^^^^^ matches above
| Your new tarball should have this md5sum:
|   3c049ed2f470d81627ea5d513a06aa15
    ^^^^^^^ matches above

Thus, I'm using an older version of wx than I had thought. The VM crashed,
I restored an older copy, I updated lmi, and I thought I had remembered to
update wx...but apparently I didn't. Well, the VM has been stable for days
(a new record) now that I nuked spice-guest-tools (that's not proof, but I
don't much believe in coincidence). I'll update wx, resave the VM, and
repeat the test. It seems reasonable to hope we've found the problem.
Thanks for your help.