many manpages cut-off in RC1

Found something interesting.. the issue itself is not consistent: it (seemingly randomly) shifs between 3 states:

* Shows 1 line : Manual page crontab(1) line ?/? (END)
* Shows the manpage, but cut-off
* Shows the complete manpage.

I;ve been able to reproduce this in my dist-upgraded devzone: try running

 while true; do man crontab;done

Any idea on this random behavior?

Another interesting datapoint:

    man {manpage} | cat

will sometimes display the entire manpage when nothing is otherwise
displayed. I wonder if TERM affects these things at all?

anilg wrote:
> Found something interesting.. the issue itself is not consistent: it
> (seemingly randomly) shifs between 3 states:
>
> * Shows 1 line : Manual page crontab(1) line ?/? (END)
> * Shows the manpage, but cut-off
> * Shows the complete manpage.
>
> I;ve been able to reproduce this in my dist-upgraded devzone: try
> running
>
> while true; do man crontab;done
>
> Any idea on this random behavior?
>
>

root@lanai:/tmp# while true; do man crontab > /tmp/$a;a=$((a+1));echo $a;done
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
^C
root@lanai:/tmp# ls -l
total 60
-rw-r--r-- 1 root root 0 May 7 12:11 0
-rw-r--r-- 1 root root 0 May 7 12:11 1
-rw-r--r-- 1 root root 5120 May 7 12:12 10
-rw-r--r-- 1 root root 5120 May 7 12:12 11
-rw-r--r-- 1 root root 5120 May 7 12:12 12
-rw-r--r-- 1 root root 0 May 7 12:12 13
-rw-r--r-- 1 root root 0 May 7 12:12 14
-rw-r--r-- 1 root root 0 May 7 12:12 15
-rw-r--r-- 1 root root 0 May 7 12:11 2
-rw-r--r-- 1 root root 0 May 7 12:11 3
-rw-r--r-- 1 root root 5120 May 7 12:11 4
-rw-r--r-- 1 root root 5120 May 7 12:11 5
-rw-r--r-- 1 root root 0 May 7 12:11 6
-rw-r--r-- 1 root root 0 May 7 12:11 7
-rw-r--r-- 1 root root 0 May 7 12:11 8
-rw-r--r-- 1 root root 12690 May 7 12:12 9

anilg wrote:
> root@lanai:/tmp# while true; do man crontab > /tmp/$a;a=$((a+1));echo $a;done
> 1
> 2
> 3
> 4
> 5
> 6
> 7
> 8
> 9
> 10
> 11
> 12
> 13
> 14
> 15
> ^C
> root@lanai:/tmp# ls -l
> total 60
> -rw-r--r-- 1 root root 0 May 7 12:11 0
> -rw-r--r-- 1 root root 0 May 7 12:11 1
> -rw-r--r-- 1 root root 5120 May 7 12:12 10
> -rw-r--r-- 1 root root 5120 May 7 12:12 11
> -rw-r--r-- 1 root root 5120 May 7 12:12 12
> -rw-r--r-- 1 root root 0 May 7 12:12 13
> -rw-r--r-- 1 root root 0 May 7 12:12 14
> -rw-r--r-- 1 root root 0 May 7 12:12 15
> -rw-r--r-- 1 root root 0 May 7 12:11 2
> -rw-r--r-- 1 root root 0 May 7 12:11 3
> -rw-r--r-- 1 root root 5120 May 7 12:11 4
> -rw-r--r-- 1 root root 5120 May 7 12:11 5
> -rw-r--r-- 1 root root 0 May 7 12:11 6
> -rw-r--r-- 1 root root 0 May 7 12:11 7
> -rw-r--r-- 1 root root 0 May 7 12:11 8
> -rw-r--r-- 1 root root 12690 May 7 12:12 9
>
>

Sweet, we have a work-around!

Seriously though it looks like one of the child processes is killed on
sigpipe while trying to write to fd=1 (stdout ... piped to parent process):

http://pastebin.com/m6a25d098

There is no exec after the fork so the code is still inside of man. The
fun part is that the truss output is pretty much the same for cases that
succeed:

http://pastebin.com/m7b51da55

-Tim