[CentOS-devel] Re: CentOS Errata

On Tuesday 04 September 2007 15:36, Karanbir Singh wrote:
> Kevan Benson wrote:
> > I just might, since I need to process all the CentOS errata anyways.
>
> interesting, why ?

A side venture of mine.  A centralized heterogeneous update notification 
system.  Instead of listing package names, it links directly to the OS's 
supplied errata page, condensing multiple packages to the single errata the 
references them.  Email me off list if you want more info than that.

Obviously, Redhat errata cover 99% of the packages listed from a CentOS 
system, but the specific CentOS packages don't show up in my program as 
Errata, but as the package name as they don't match a known package from 
Redhat.

I would much rather list official CentOS errata info for those packages than 
skip directly to the Redhat errata page the package was modified from, for 
multiple reasons (and most of them aren't self-serving, I promise).

> > I could whip up some MySQL schema that stores the relevant errata info
> > CentOS publishes and a CGI or PHP script of some sort could display it as
> > needed, RSS or HTML, even take care of the mailing when a new errata is
> > submitted (or after X days and new errata exist).  That would kill two
> > (or three!) birds with one stone.
>
> considering we are unlikely to have millions of announcements, I'd pass
> on the database. Surely, a flat file should do the job nicely. Mailings
> for CentOS-5 come from the buildsystem, and there is a fair bit of
> automation with CentOS-4/3 announcements as well.

Well, it mailing are already automated, that does take away some of the 
impetus to implement this.  Although there is still the added benefit that a 
the web based errata archive (see below) and the RSS will necessarily always 
be in sync, adding/modifying an errata entry would be simple, and the display 
of errata info and RSS of errata info could both be 60-or-so line scripts.

> > Although, if you already have something lined up for the RSS, it might
> > make sense to just generate the HTML for the errata and put it up, so
> > there's at least something to look at for archives.>
>
> I seem to be missing the point here, if the rss feed exists, why do we
> need to further process this into html ?

There may not be a need, but it is very nice to be able to go to your OS 
vendor's web page and search for anything that might affect your system, or 
for info regarding a specific package and actually have information 
available.

I'll admit, it does fit my above stated goal, but that's not the only reason I 
think it's worthwhile.  Every other enterprise OS has their errata listed 
online.  CentOS seems to be in the somewhat unique situation of having an 
upstream provider that has most the errata listed, so there's been less of a 
drive for this.

> >  I'll be processing the old errata
> > anyways, not hard to pull it all out and stick it in some directory
> > structure while I'm at it.  That does lead to a need for someone to
> > maintain it though.
>
> again, i seem to be missing the point here. Why do you need to do this ?
> just use the yum metadata for everything and you should be sorted.

Yeah, I'll admit my needs are somewhat unique. :)

-- 
- Kevan Benson
- A-1 Networks