Sergey Poznyakoff wrote:
Yeah, I know. I decided to retain lzma for a while... There is no
harm in distributing an extra compressed variant, isn't it?
I would agree with you if it were some other compressed variant, but not
lzma-alone, because while other formats have the means to detect errors,
(and lzip is even able to repair some of them), lzma-alone silently
produces corrupt data too easily. Believe me that I suffer every time I
see a friend using it.
Take for example the file mailutils-2.2.tar.lzma:
$ ls -l mailutils-2.2.tar.*
-rw-r--r-- 1 antonio users 3092647 2010-09-10 17:31 mailutils-2.2.tar.bz2
-rw-r--r-- 1 antonio users 2067042 2010-09-10 17:50 mailutils-2.2.tar.lzma
Now change byte 26883 of mailutils-2.2.tar.lzma from 0x24 to 0x1C and
test it with lzma or xz:
$ lzma -tv mailutils-2.2-bad.tar.lzma
mailutils-2.2-bad.tar.lzma: decoded successfully
$ xz -tv mailutils-2.2-bad.tar.lzma
mailutils-2.2-bad.tar.lzma (1/1)
100.0 % 2018.6 KiB / 19.5 MiB = 0.101
Seems good, but compare it with the output of bzip2:
$ bzip2 -kd mailutils-2.2.tar.bz2
$ lzma -cd mailutils-2.2-bad.tar.lzma | diff -au mailutils-2.2.tar -
--- mailutils-2.2.tar 2010-09-10 17:31:06.000000000 +0200
+++ - 2010-09-12 21:20:01.000000000 +0200
@@ -3398,7 +3398,7 @@
| mt-* \
| msp430-* \
| nios-* | nios2-* \
- | none-* | np1-* | ns16k-* | ns32k-* \
+ | none-* | np1-* le | sh-* lens32k-* \
| orion-* \
| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* | ppcbe-* \
I expect this to produce at least some spurious bug reports.
Best regards,
Antonio.
_______________________________________________
Bug-mailutils mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/bug-mailutils
