Re: Re : new version of gpgme (0.4.5) breaks signature checking

[Albrecht Dreß <albrecht dress arcor de>]

Am 18.03.04 22:24 schrieb(en) Jean-Luc Coulon (f5ibh):
> I've done this test and got 2 log file : one for libgpgme 0.4.3 and the  
> other with 0.4.5. The balsa executable is the same and from the cvs 2-0  
> with your "anti-colorisation" [tm] patch (BTW, the orange colour when a  
> signature is invalid is really ugly ;-) ).

That should be the same colour as used for the "invalid address" label  
(which you can change in the prefs). Seems to be a semi-optimal solution,  
though...

Now for the logs - thanks a lot! The interesting thing from the 0.4.5 one  
is that the RFC2440 message seems to work fine (is that true?), at least  
the data io is perfect. I must admit that I see what's going on in gpgme  
with the '3156 signed one (the second in the log), but I can't tell you  
why it fails. The problem is at the very beginning of processing that  
message:

[snip]
posix-io.c:156: set notification for fd 33
posix-io.c:156: set notification for fd 34
posix-io.c:134: closing fd 30
posix-io.c:134: closing fd 31
posix-io.c:134: closing fd 33
posix-io.c:328: gpgme:select on [ r29 w32 w34 ]
posix-io.c:374: select OK [ w32 w34 ]
posix-io.c:134: closing fd 32
wait.c:159: setting fd 32 (item=0x8409610) done
[snip]

After the "select ok" (telling us that handles 32 and 34 are ready for  
writing) #32 sould not be closed, but filled with the signature... From  
the other (0.4.3) log I see that it's present, and the "wait" below  
indicates that the data object has been created correctly.

Before I pass that to the gpgme list, it might be a good idea to check a  
few things:

- I guess you built balsa with threading (pthread) support? Is the binary  
correctly linked against libgpgme-pthread (try "ldd balsa")? It might be  
worth a try to build balsa *without* threading and see if you get the same  
effect with 0.4.5;

- Did you build gpgme-0.4.5 yourself, or is it from a bin package? The  
only important change apart from a few bug fixes between 0.4.3/0.4.4 and  
0.4.5 was the introduction of large file support (64 bit length), which  
shouldn't make any difference. So if it's a bin package, it might be  
interesting to see what happens if you build gpgme (or gpgme and libgpg- 
error) from sources; it's pretty easy, just "./configure; make all  
install".

Cheers, Albrecht.


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Albrecht Dreß  -  Johanna-Kirchner-Straße 13  -  D-53123 Bonn (Germany)
       Phone (+49) 228 6199571  -  mailto:albrecht.dress@arcor.de
_________________________________________________________________________

PGP signature