Re: [Bug-cssc] bug-CSSC post from address@hidden requires approval

[Joerg Schilling]

James Youngman <address@hidden> wrote:

> > Well, opengrok includes support for sccs so there is even a GUI for sccs.
> > Why don't you use it anymore? It still seems to be based on the most robust
> > file format specification and I know nobody who did ever loss data with SCCS
> > during the past 30+ years.
>
> To be fair though, not corrupting data is the absolutely lowest
> possible requirement for a configuration management system.

I've seen three projects on Berlios losing all files because they did use svn 
when svn was a bit younger than now. There was no way to recover.

> I've certainly seen _incorrectly updated_ SCCS files.   Cases of
> deletions or insertions being doubled where repeated sequences of get
> -e / delta straddle large backward clock changes.   However, this was
> on Dynix, just before IBM bought Sequent.  So I hardly think that
> historic defect on a now-obsolete version of SCCS is going to be
> material these days.

Well, SCCS normally does not care about time stamps. This is why non-monotonic 
time stamps could make it into the history files. Note that having something 
like ntp running was not usual until recently, so a reboot could cause the 
system clock to be wrong for a longer time.

SCCS should at least check whether a new delta is created with a monotonicly 
increasing time. In fact, this was done in 1980 already but there is only a 
message and people may have ignored it.

Jörg

-- 
 EMail:address@hidden (home) Jörg Schilling D-13353 Berlin
       address@hidden                (uni)  
       address@hidden (work) Blog: http://schily.blogspot.com/
 URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily