Re: [lmi] wx_test: output strings truncated to one character

[Vadim Zeitlin]

On Wed, 10 Feb 2016 15:20:07 +0000 Greg Chicares <address@hidden> wrote:

GC> I reran 'install_wx.make' with '--disable-precomp-headers' but without
GC> any "debug" configure flag. This rebuilds and reinstalls wx completely.
GC> I didn't do anything with wxPdfDoc, and didn't recompile lmi. Result:
GC> 
GC> /opt/lmi/bin[0]$./wx_test.exe --ash_nazg --data_path=/opt/lmi/data
GC> pure virtual method called

 So this only happens when optimizations are enabled :-(

GC> i: started
GC> In file ../include/wx/msw/private.h at line 834: 'UnregisterClass' failed 
with error 0x00000584 (class still has open windows.).
GC> /opt/lmi/bin[1]$
GC> 
GC> So it once again terminates abnormally (with a messagebox), which it
GC> didn't do with the '--enable-debug' wx build; but it does not segfault
GC> as it formerly did.

 I think the last problem (UnregisterClass() and message box) could be
"just" a consequence of an imperfectly reproducible test failure. This is
still important to fix, of course, but it doesn't bother me as much as the
pure virtual call on startup or the segfault at the end.

GC> I reran 'install_wx.make', this time with both '--disable-precomp-headers'
GC> and '--enable-debug_info', then reran the GUI test. Same outcome as above.

 This is reassuring. If you still have this build, could you please try to
break on the pure virtual call under gdb to check if you see any more
information about where it happens?

GC> >  All this is obviously very worrisome because it's a clear regression
GC> > compared to MinGW 3.4, but unfortunately I just can't do much about it 
from
GC> > here as long as I can't even reproduce the bug :-(
GC> 
GC> At this point I'm no longer sure: which anomaly are we talking about?
GC> The segfault that I can no longer reproduce? Or the "pure virtual" and
GC> "'UnregisterClass' failed" messages?

 For me "pure virtual" is the most mysterious problem. The next one is the
segfault, but if it went completely away, perhaps we can explain it as some
build artefact. UnregisterClass() and the message box look like a normal
bug and I'll try debugging them later, once you confirm that they still
happen with the latest wx.

GC> Anyway, I think I've travelled this particular path to the end, and now
GC> it's time to update to wx HEAD.

 Thanks in advance!
VZ