Re: script(1) bug?

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

On Fri, Aug 28, 2026 at 13:51:31 +0300, Valery Ushakov wrote:

> indeed: consider
> 
>   runme() {
>     rm typescript
>     script -q -f -c 'echo foo; sleep 10;' < /dev/null > /dev/null
>     ls -l typescript
>     hexdump -C typescript
>   }
> 
> 
> And then you race:
> 
> $ runme
> -rw-r--r--  1 uwe  uwe  5 Aug 28 13:44 typescript
> 00000000  66 6f 6f 0d 0a                                    |foo..|
> 00000005
> $ runme
> -rw-r--r--  1 uwe  uwe  0 Aug 28 13:46 typescript
> $runme
> -rw-r--r--  1 uwe  uwe  0 Aug 28 13:44 typescript
> 00000000  66 6f 6f 0d 0a                                    |foo..|
> 00000005

At the end of main(), if read(2) from stdin fails or returns EOF, the
script(1) should not exit and should do sigsuspend(2).  Shouldn't be
too hard to fix, but needs someone with tuits...

-uwe