On 20/09/2022 06:46, Tom Tanner wrote:
On 19/09/2022 18:28, Dominique Dumont wrote:
Hi
Thanks, Duncan, for the detail analysis.
On Monday, 19 September 2022 07:59:30 CEST Duncan wrote:
I'm compiling using gcc-12.2.0 on gentoo ~amd64.
Looks like iconv_t is defined by /usr/include/iconv.h, which is from
glibc
so it's pretty basic. Today's update did include a glibc update, to
glibc-2.35-r10 (a gentoo-revision update from the previous -r8, seems I
skipped -r9), but I did confirm that my previously installed git
7771e5655
still built against the new glibc, as did the immediately pre-culprit
commit 503708583, so indeed the culprit is 18f92b133.
Thomas, this failure occur when pan is compiled without gmime crypto.
The commit pointed by Duncan updates pan/usenet-utils/gpg.h with:
#ifndef _HAVE_GPGDEFS_H
#define _HAVE_GPGDEFS_H
+#include <config.h>
+
+#ifdef HAVE_GMIME_CRYPTO
+
#include <gmime/gmime.h>
#include <map>
#include <vector>
-#ifdef HAVE_GMIME_CRYPTO
-
Was there any reason to move up the #ifdef line ?
All the best
_______________________________________________
Pan-users mailing list
Pan-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/pan-users
Hmm. This looks like it's a transitive include issue. Let me check
Yes. It looks like a more appropriate fix would be to add the gmime.h
include to mime-utils.h, with this patch:
diff --git a/pan/usenet-utils/mime-utils.h b/pan/usenet-utils/mime-utils.h
index a4cc647..d7b4e16 100644
--- a/pan/usenet-utils/mime-utils.h
+++ b/pan/usenet-utils/mime-utils.h
@@ -23,6 +23,7 @@
#include <config.h>
#include <vector>
#include <glib.h>
+#include <gmime/gmime.h>
#include <gmime/gmime-filter.h>
#include <gmime/gmime-stream.h>
#include <gmime/gmime-message.h>
sorry for the hassle
(I am being very pedantic here, sorry...)
_______________________________________________
Pan-users mailing list
Pan-users@nongnu.org
https://lists.nongnu.org/mailman/listinfo/pan-users