Re: [dev] [sbase] about audit

From: FRIGN <>
Date: Thu, 1 Sep 2016 17:56:07 +0200

On Thu, 1 Sep 2016 12:34:12 -0300
Marc Collin <> wrote:

Hey Marc,

> Yeah, I'm not complaining about the audits, I remember they found
> many bugs! It was just a random thing that came to me - normally a
> person that makes a mistake will overlook the mistake. But someone
> from the outside should spot it more easily.
> I don't even know if this is true, but I think it is based on self
> experience. And I think the message was good, since now Ali H. Fardan
> decided to go through the code too.
> In the end everyone wins :)

the fault was a bit at me because the audit-patches were so big and
nobody would be arsed to check the mall.
However, nobody stops you from subscribing to hackers_AT_ and receiving
all the commits. Open Source lives from many eyes checking changes.

> Since we're talking about sbase already in kind of meta way, I'll post
> a question here instead of a new email.
> sbase is basically ready, right? The few missing tools are not yet
> applied, but were sent to the ML by maandree some months ago (patch,
> diff and others). Should we expect a release soon? I'm excited :)

Oh well, this is a difficult topic. Sbase has received a lot of work
and it works better in some respects than the GNU coreutils. Aside from
sed, ed and grep the tools are well-audited and I am confident they can
be used in everyday life.
However, there are still some things to do, but it's really not
something preventing a release. Up to this point, we have not gotten
around for doing a release because there is always something that can
get in the way.
Most prominently, nobody wants to make a release only to get a bug
report the next day for a critical problem in some tool. We tested the
sbase tools extensively, but some bugs just show up after heavy usage
or deployment.

In some respects, sbase is almost too ambitious for the given manpower.
An important point for instance is the UTF-8/Unicode stuff. I expanded
libutf to support proper case conversions, but we are still lacking in
detecting grapheme clusters and multi-codepoint case conversions.
This all depends on my redevelopment of libutf that is currently in the
works, but given some personal things I have not gotten around to it in
the last few months and thus development kinda halted.



Received on Thu Sep 01 2016 - 17:56:07 CEST

This archive was generated by hypermail 2.3.0 : Thu Sep 01 2016 - 18:00:27 CEST