Hi!

On Fri, 2015-04-17 at 17:20:00 -0400, Jim Carroll wrote:
> > > We'd prefer to use inetutils-ping as it is included by
> > > default on most debian systems. And that makes it easily
> > > available to Docker images built on debian.
> >
> > (Just curious, but what do you mean with "included by default"? It's
> > Priority: extra, while iputils-ping is Priority: important.)
> 
> I think this is my lack of knowledge about Debian's package system. What I
> mean to say was that I discovered that most of the Debian system's I'd
> logged into seemed to already have the 'inetutils-ping' version of ping
> installed. I had assumed this correlated to some sort of default
> installation since I had not specifically requested it in my builds.
> 
> Once place where this really caught us was our use of Docker. All the Debian
> images (which sit at the base of tons of other images) all seem to include
> the 'inetutils-ping' version of ping, rather than the 'iputils-ping' editon.
> Obviously it's not a big deal to just define our own images and make a
> substitution -- but its extra maintenance and takes us away from the
> standard blessed images available from the Docker repos.

Ah, ok. I guess that for some reason all those sites prefer inetutils'
implementation. My guess would be because it is more portable, but I
wouldn't really know.

> I'll spend some time reading up on the priority attributes of debian
> packages to make sure any future communication is more precise.

You can check this out:

<https://www.debian.org/doc/debian-policy/ch-archive.html#s-priorities>

> If you don't mind, I have one followup question regarding a comment you
> made:
> 
> > Actually the problem is that stdout when redirected to a file switches
> > to fully buffered output, contrary to when writing to a terminal where
> > it does line buffering.
> 
> Could you clarify this a bit more? Who is enabling file buffering? Is it the
> shell? Something in the plumbing of IPC/pipes?  Something in the kernel? Any
> details would be greatly appreciated.

Those messages are printed through stdio streams, those are initialized
and part of glibc, so that's who decides about the buffering.

I just checked and this seems to be briefly documented in «man stdio».

Thanks,
Guillem


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to