Bug#438385: NMU awardeco #438385: fails on 64-bit platforms
Hello,
I'm ready for a NMU, attached I put the diff.
Now I'll upload the package in delayed.
the patch contain:
- fix on 64-bit architecture: use C99 "fixed"-bit integer
- fix of undeclared function: a potential bug, which can
give problem on some architectures.
Considering the package pourpose, maybe Architecture: "any"
is to wide.
- fix a return of local stack variables
PS: only the first is a rc bug (#438385),
but the second adn the third are potential rc or important bugs.
ciao
cate
diff -u awardeco-0.2/debian/changelog awardeco-0.2/debian/changelog
--- awardeco-0.2/debian/changelog
+++ awardeco-0.2/debian/changelog
@@ -1,3 +1,11 @@
+awardeco (0.2-2.1) unstable; urgency=low
+
+ * Non-Maintainer Upload at BSP in Zurich: fix rc bug
+ * Use the C99 bit length integer, to be safe on various archs
+ * fix also headers inclusion (memcpy: , exit:
+
+ -- Giacomo Catenazzi <[EMAIL PROTECTED]> Sat, 12 Jan 2008 20:07:12 +0100
+
awardeco (0.2-2) unstable; urgency=low
* Fix FTBFS on GNU/kFreeBSD (Closes: #414235).
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/30_fixeduint.patch
+++ awardeco-0.2/debian/patches/30_fixeduint.patch
@@ -0,0 +1,17 @@
+diff -Nur awardeco-0.2/src/awardeco.h awardeco-0.2.new/src/awardeco.h
+--- awardeco-0.2/src/awardeco.h 2006-08-05 11:38:03.0 +0200
awardeco-0.2.new/src/awardeco.h 2008-01-16 08:12:10.0 +0100
+@@ -10,10 +10,11 @@
+ #define AWARDECO_H 1
+
+ #include
++#include
+
+ typedef unsigned char byte;
+-typedef unsigned short word;
+-typedef unsigned long dword;
++typedef uint16_t word;
++typedef uint32_t dword;
+
+ #define Xtract 0x10
+ #define List 0x11
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/40_fixheader.patch
+++ awardeco-0.2/debian/patches/40_fixheader.patch
@@ -0,0 +1,12 @@
+diff -Nur awardeco-0.2/src/awardfnc.c awardeco-0.2.new/src/awardfnc.c
+--- awardeco-0.2/src/awardfnc.c 2006-08-05 11:38:03.0 +0200
awardeco-0.2.new/src/awardfnc.c 2008-01-16 08:25:31.0 +0100
+@@ -10,6 +10,8 @@
+ #define AWARDFUNC_H 1
+
+ #include
++#include
++#include
+
+ typedef unsigned char byte;
+ typedef unsigned short word;
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/40_fixwarnings.patch
+++ awardeco-0.2/debian/patches/40_fixwarnings.patch
@@ -0,0 +1,21 @@
+diff -Nur awardeco-0.2/src/awardfnc.c awardeco-0.2.new/src/awardfnc.c
+--- awardeco-0.2/src/awardfnc.c 2008-01-16 08:28:09.0 +0100
awardeco-0.2.new/src/awardfnc.c 2008-01-16 08:30:41.0 +0100
+@@ -153,7 +153,7 @@
+
+ byte *GetFullDate(byte *mon, byte *day, byte *year)
+ {
+- byte *Months[]={"",
++ static byte const*const Months[]={"",
+ "January",
+ "February",
+ "March",
+@@ -166,7 +166,7 @@
+ "October",
+ "November",
+ "December"};
+- byte Buf[20];
++ static byte Buf[20]; // Warning: we return this buffer, so place it in heap not on stack!
+ sprintf(Buf,"%2.2s",year);
+
+ if((atoi(Buf)>=0) && (atoi(Buf)<70)) sprintf(Buf,"%.2s %s %s%.2s",day,Months[atoi(mon)],"20",year);
Bug#438385: NMU awardeco #438385: fails on 64-bit platforms
Hi, On Wed, Jan 16, 2008 at 08:51:48AM +0100, Giacomo Catenazzi wrote: > I'm ready for a NMU, attached I put the diff. > Now I'll upload the package in delayed. Thanks, feel free to upload, but please see comments below. > Considering the package pourpose, maybe Architecture: "any" > is to wide. Nope, don't think so. The package is supposed to work everywhere. > diff -u awardeco-0.2/debian/changelog awardeco-0.2/debian/changelog > --- awardeco-0.2/debian/changelog > +++ awardeco-0.2/debian/changelog > @@ -1,3 +1,11 @@ > +awardeco (0.2-2.1) unstable; urgency=low > + > + * Non-Maintainer Upload at BSP in Zurich: fix rc bug > + * Use the C99 bit length integer, to be safe on various archs Please add "(Closes: #438385)" here. Also, please end all changelog lines with a full stop and start sentences with capital letters for consistency. > --- awardeco-0.2.orig/debian/patches/30_fixeduint.patch > +++ awardeco-0.2/debian/patches/30_fixeduint.patch > @@ -0,0 +1,17 @@ > +diff -Nur awardeco-0.2/src/awardeco.h awardeco-0.2.new/src/awardeco.h > +--- awardeco-0.2/src/awardeco.h 2006-08-05 11:38:03.0 +0200 > awardeco-0.2.new/src/awardeco.h 2008-01-16 08:12:10.0 +0100 > +@@ -10,10 +10,11 @@ > + #define AWARDECO_H 1 > + > + #include > ++#include > + > + typedef unsigned char byte; Maybe make this uint8_t then, too, for consistency? Yes, it's more cosmetical, but still. Cheers, Uwe. -- http://www.hermann-uwe.de | http://www.holsham-traders.de http://www.crazy-hacks.org | http://www.unmaintained-free-software.org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#438385: NMU awardeco #438385: fails on 64-bit platforms
diff -u awardeco-0.2/debian/changelog awardeco-0.2/debian/changelog --- awardeco-0.2/debian/changelog +++ awardeco-0.2/debian/changelog @@ -1,3 +1,11 @@ +awardeco (0.2-2.1) unstable; urgency=low + + * Non-Maintainer Upload at BSP in Zurich: fix rc bug + * Use the C99 bit length integer, to be safe on various archs Please add "(Closes: #438385)" here. Also, please end all changelog lines with a full stop and start sentences with capital letters for consistency. oops. Ok. the first was a cut-copy error trying cdbs-edit-patch (nice tools!). Never realized about cosmetic consistency, I'll fix also that. --- awardeco-0.2.orig/debian/patches/30_fixeduint.patch +++ awardeco-0.2/debian/patches/30_fixeduint.patch @@ -0,0 +1,17 @@ +diff -Nur awardeco-0.2/src/awardeco.h awardeco-0.2.new/src/awardeco.h +--- awardeco-0.2/src/awardeco.h2006-08-05 11:38:03.0 +0200 awardeco-0.2.new/src/awardeco.h2008-01-16 08:12:10.0 +0100 +@@ -10,10 +10,11 @@ + #define AWARDECO_H 1 + + #include ++#include + + typedef unsigned char byte; Maybe make this uint8_t then, too, for consistency? Yes, it's more cosmetical, but still. yes. This change doesn't harm: in C, char is a byte and in POSIX, char is 8 bit. So I take a smaller NMU. But reading the code it seems ok to do this conversion. I've some other minor stylistic C correction, i.e. "0x;" is wrong, it should have a "u" postfix. (and "ul" if we care about real C, but anyway POSIX int is 32-bit or more.), but for this I think I (or you) should contact upstream author. ciao cate -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#438385: NMU awardeco #438385: fails on 64-bit platforms
Hello,
Now I upload (still to DELAYED) the new version,
diff attached.
I have fixed the problems you pointed in last mail,
and I noticed I forgot one part of the uintXX_t
patch.
BTW:
lintian give two warning (on your and on this NMU)
- policy version
- empty /usr/bin
I see you install the binary on /bin and not
on /usr/bin. Is really what you want?
Do you know why pool
http://ftp.debian.org/debian/pool/main/a/awardeco/
contains only i386 and amd64 ?
But the package should be already build and installed on all archs:
http://people.debian.org/~igloo/status.php?packages=awardeco
ciao
cate
diff -u awardeco-0.2/debian/changelog awardeco-0.2/debian/changelog
--- awardeco-0.2/debian/changelog
+++ awardeco-0.2/debian/changelog
@@ -1,3 +1,12 @@
+awardeco (0.2-2.1) unstable; urgency=low
+
+ * Non-Maintainer Upload at BSP in Zurich: fix rc bug.
+ * Use the C99 bit length integer, to be safe on various
+architectures (Closes: #438385).
+ * Fix also headers inclusion (memcpy: , exit: ).
+
+ -- Giacomo Catenazzi <[EMAIL PROTECTED]> Thu, 17 Jan 2008 08:17:08 +0100
+
awardeco (0.2-2) unstable; urgency=low
* Fix FTBFS on GNU/kFreeBSD (Closes: #414235).
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/30_fixeduint.patch
+++ awardeco-0.2/debian/patches/30_fixeduint.patch
@@ -0,0 +1,36 @@
+diff -Nur awardeco-0.2/src/awardeco.h awardeco-0.2.new/src/awardeco.h
+--- awardeco-0.2/src/awardeco.h 2006-08-05 11:38:03.0 +0200
awardeco-0.2.new/src/awardeco.h 2008-01-17 08:25:23.0 +0100
+@@ -10,10 +10,11 @@
+ #define AWARDECO_H 1
+
+ #include
++#include
+
+-typedef unsigned char byte;
+-typedef unsigned short word;
+-typedef unsigned long dword;
++typedef uint8_t byte;
++typedef uint16_t word;
++typedef uint32_t dword;
+
+ #define Xtract 0x10
+ #define List 0x11
+diff -Nur awardeco-0.2/src/awardfnc.c awardeco-0.2.new/src/awardfnc.c
+--- awardeco-0.2/src/awardfnc.c 2006-08-05 11:38:03.0 +0200
awardeco-0.2.new/src/awardfnc.c 2008-01-17 08:26:24.0 +0100
+@@ -10,10 +10,11 @@
+ #define AWARDFUNC_H 1
+
+ #include
++#include
+
+-typedef unsigned char byte;
+-typedef unsigned short word;
+-typedef unsigned long dword;
++typedef uint8_t byte;
++typedef uint16_t word;
++typedef uint32_t dword;
+
+ #define Xtract 0x10
+ #define List 0x11
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/40_fixheader.patch
+++ awardeco-0.2/debian/patches/40_fixheader.patch
@@ -0,0 +1,12 @@
+diff -Nur awardeco-0.2/src/awardfnc.c awardeco-0.2.new/src/awardfnc.c
+--- awardeco-0.2/src/awardfnc.c 2008-01-17 08:26:37.0 +0100
awardeco-0.2.new/src/awardfnc.c 2008-01-17 08:27:05.0 +0100
+@@ -11,6 +11,8 @@
+
+ #include
+ #include
++#include
++#include
+
+ typedef uint8_t byte;
+ typedef uint16_t word;
only in patch2:
unchanged:
--- awardeco-0.2.orig/debian/patches/40_fixwarnings.patch
+++ awardeco-0.2/debian/patches/40_fixwarnings.patch
@@ -0,0 +1,21 @@
+diff -Nur awardeco-0.2/src/awardfnc.c awardeco-0.2.new/src/awardfnc.c
+--- awardeco-0.2/src/awardfnc.c 2008-01-16 08:28:09.0 +0100
awardeco-0.2.new/src/awardfnc.c 2008-01-16 08:30:41.0 +0100
+@@ -153,7 +153,7 @@
+
+ byte *GetFullDate(byte *mon, byte *day, byte *year)
+ {
+- byte *Months[]={"",
++ static byte const*const Months[]={"",
+ "January",
+ "February",
+ "March",
+@@ -166,7 +166,7 @@
+ "October",
+ "November",
+ "December"};
+- byte Buf[20];
++ static byte Buf[20]; // Warning: we return this buffer, so place it in heap not on stack!
+ sprintf(Buf,"%2.2s",year);
+
+ if((atoi(Buf)>=0) && (atoi(Buf)<70)) sprintf(Buf,"%.2s %s %s%.2s",day,Months[atoi(mon)],"20",year);
Bug#438385: NMU awardeco #438385: fails on 64-bit platforms
On Thu, Jan 17, 2008 at 08:50:29AM +0100, Giacomo Catenazzi wrote: > Now I upload (still to DELAYED) the new version, > diff attached. Thanks! > lintian give two warning (on your and on this NMU) > - policy version That's non-critical, I'll update it in the next upload. > - empty /usr/bin > I see you install the binary on /bin and not > on /usr/bin. Is really what you want? Oops, no. That's a bug indeed. I'll fix it in the next upload. > Do you know why pool > http://ftp.debian.org/debian/pool/main/a/awardeco/ > contains only i386 and amd64 ? > But the package should be already build and installed on all archs: > http://people.debian.org/~igloo/status.php?packages=awardeco Hm, no idea. It does build fine everywhere though, see http://buildd.debian.org/pkg.cgi?pkg=awardeco Uwe. -- http://www.hermann-uwe.de | http://www.holsham-traders.de http://www.crazy-hacks.org | http://www.unmaintained-free-software.org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
