>I think this (and many of the patches you sent) it is just *your* problems
>with a broken compiler. The Standard define non-local non-initialized
>object to be pre-initialized with zero.
Where is this standard written? I don't know of neither msvc or gcc to
initialize variables with zero, at l
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Situation:
>
> - Win32 [release or] debug build (yep, I use MSVC2005 project files, NOT
> the makefile, but that does not make any real difference here)
> - link to the DLL version of the OpenSSL library (0.9.9 tested, will
> VERY probably happe
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> If you use -SI (-PI / -XI / -FI / -CI / -WI ) with the
> optional [:port] section included to specify a non-standard port for the
> interface address(es), you're SOL as the code calls the appropriate
> routine but does not extract the decoded
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Hi,
>
> There's still a bug lurking in the code, but now for a fault situation
> only fortunately, far as I can tell.
>
> How to reproduce:
>
> - use Outlook Express
> - create account
> - FORGET to check the checkmark 'Outgoing (SMTP) server r
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Minor memory leaks were detected when certain errors had been reported.
>
> Best regards,
>
> Ger
>
>
> PS: also in this fix: zeroed static data because MSVC2005 does NOT zero
> static data unless specifically instructed to do so (in analysis
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Hi,
>
> While hunting down another issue (to be posted shortly), I ran into this
> one. Shouldn't the '&&' be replaced with a '||' for security reasons in
> error situations? Otherwise error situations in [ONLY] ONE of the calls
> (due to wrong
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Situation:
>
> - Win32 [release or] debug build (yep, I use MSVC2005 project files, NOT
> the makefile, but that does not make any real difference here)
> - link to the DLL version of the OpenSSL library (0.9.9 tested, will
> VERY probably happe
On Tue, 20 Nov 2007, Gerrit E.G. Hobbelt wrote:
> When you've got that situation where those empty directories don't all
> exist (just like the documentation is warning you about), XMAIL will
> sometimes treat such a situation not as the severe error it should be,
> but instead ignores the fail
Hi,
There's still a bug lurking in the code, but now for a fault situation
only fortunately, far as I can tell.
How to reproduce:
- use Outlook Express
- create account
- FORGET to check the checkmark 'Outgoing (SMTP) server requires
authentication'
- start XMAIL server and have SMTP-after-POP
Minor memory leaks were detected when certain errors had been reported.
Best regards,
Ger
PS: also in this fix: zeroed static data because MSVC2005 does NOT zero
static data unless specifically instructed to do so (in analysis debug
builds at least), leading to random data in these structs an
If you use -SI (-PI / -XI / -FI / -CI / -WI ) with the
optional [:port] section included to specify a non-standard port for the
interface address(es), you're SOL as the code calls the appropriate
routine but does not extract the decoded port for further use, nor does
it mix well with any atte
Hi,
While hunting down another issue (to be posted shortly), I ran into this
one. Shouldn't the '&&' be replaced with a '||' for security reasons in
error situations? Otherwise error situations in [ONLY] ONE of the calls
(due to wrong setup or other gimmicks) would INCORRECTLY and UNDESIRABLY
Situation:
- Win32 [release or] debug build (yep, I use MSVC2005 project files, NOT
the makefile, but that does not make any real difference here)
- link to the DLL version of the OpenSSL library (0.9.9 tested, will
VERY probably happen with other versions too, as long as you use OpenSSL
DLLs!)
When you've got that situation where those empty directories don't all
exist (just like the documentation is warning you about), XMAIL will
sometimes treat such a situation not as the severe error it should be,
but instead ignores the failure.
This has to do with the fact that the OS error is n
Davide Libenzi wrote:
> Thanks, applied.
>
>
You're welcome ;-)
There's a few more minor tidbits, which I'll be posting diffs for in the
next few weeks (got to separate them out for easier processing).
Cheers,
Ger
-
To unsubscribe from this list: send the line "unsubscribe xmail" in
the bod
On Mon, 19 Nov 2007, Gerrit E.G. Hobbelt wrote:
> Haven't checked the complete docs yet, so there may remain some minor
> typos. Anyway, I hope this is useful.
Thanks, applied.
- Davide
-
To unsubscribe from this list: send the line "unsubscribe xmail" in
the body of a message to [EMAIL PRO
Haven't checked the complete docs yet, so there may remain some minor
typos. Anyway, I hope this is useful.
Best regards,
Ger
Also no more warnings when running cvpod.pl *.cvpod in the docs dir.
diff of the .pod files follows:
--
17 matches
Mail list logo