Paul Miller <[EMAIL PROTECTED]> writes:
> How can I install hamm from scratch (I just installed 1.3.1
> yesterday)? I downloaded ncurses3.4-*.deb, is there anything else I
> need which isn't where is should be (unstable dirs)?
I apologize for bringing up what I expect is an old thread but I have
been away for a month and off the list. I too have upgraded to hamm
and many, many things are broken. I have the problems with
dependencies on ncurses3.4 which does not seem to be available. I am
happy to download it by hand and install it but I don't know where to
find it. I don't see it listed in the ls-lR.gz file from the top of
the Debian distribution.
I also have the problem with libc5 and libc5-altdev having
inconsistent dependencies (libc5-altdev requires version 5.4.23-4 of
libc5 but the version in the same tree is later). I think I know how
to fix that one.
I have a problem with the fvwm family of packages related to the
menu-methods. Then end of an Install within dselect produces errors
like
Do you wish to delete the installed package (.deb) files? [y]:
/usr/lib/menu/sced: x11: command not found
Cannot open file /etc/X11/fvwm95//system.fvwm2rc95
/etc/menu-methods//fvwm95: Aborting
The xemacs in the 19.15-3.1 package starts up, tries to load its
initialization files, then dies and dumps core. Perhaps this is
related to the updating of the menus.
The most frustrating error for me though is the problem with the
the function ecvt in libc6. I reported an error with that over a
month ago. It is not a subtle error in that *every* result from that
function is wrong. It is not that it occasionally produces a somewhat
incorrect answer - it never produces the correct answer. For example
bash-2.00$ cat ecvt_test.c
#include <stdio.h>
#include <stdlib.h>
void main(int argc, char **argv)
{
int decpt, sign; char * result;
result = ecvt((double) 1234.567, (size_t) 7, &decpt, &sign);
printf("result is %s, decpt is %d, sign is %d\n", result, decpt, sign);
result = ecvt((double) -987.654321, (size_t) 7, &decpt, &sign);
printf("result is %s, decpt is %d, sign is %d\n", result, decpt, sign);
result = ecvt((double) 0, (size_t) 7, &decpt, &sign);
printf("result is %s, decpt is %d, sign is %d\n", result, decpt, sign);
}
bash-2.00$ gcc ecvt_test.c -o ecvt_test
bash-2.00$ ./ecvt_test
result is 11234.567, decpt is 4, sign is 0
result is 9987.6543, decpt is 3, sign is 1
result is 00.000000, decpt is 1, sign is 0
bash-2.00$ /usr/bin/i486-linuxlibc1-gcc ecvt_test.c -o ecvt_test
bash-2.00$ ./ecvt_test
result is 1234567, decpt is 4, sign is 0
result is 9876543, decpt is 3, sign is 1
result is 0000000, decpt is 0, sign is 0
I am trying to work on a large system written in C and I can't do
anything with it under libc6. The fact that the decpt value is wrong
for an argument of zero causes the program to crash and one encounters
the number zero fairly frequently.
There are enough things that don't work with the altdev packages that
I can't use them to compile this system. It appears my only option is
to back out all the libc6 stuff and try to go back to libc5 and I
expect that will be a major hassle.
If this problem with libc6 is from the GNU libc2 sources, I don't see
how it would have gotten out of a testing phase. Any regression test
is going to show that you are getting different results.
I apologize for venting. I know Debian is a volunteer organization
and I appreciate the work of the volunteers. I am just frustrated
right now that so many things are broken and it is not obvious to me
how to fix them. Suggestions welcome.
--
Douglas Bates [EMAIL PROTECTED]
Statistics Department 608/262-2598
University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
[EMAIL PROTECTED] .
Trouble? e-mail to [EMAIL PROTECTED] .
