RPM Community Forums

Mailing List Message of <rpm-devel>

Re: XAR preliminaries: eliminate lead and signature header.

From: Jeff Johnson <n3npq@mac.com>
Date: Tue 31 Jul 2007 - 19:19:55 CEST
Message-Id: <365F789B-739F-4B65-BFAE-C094B8007187@mac.com>

On Jul 30, 2007, at 1:53 PM, Jeff Johnson wrote:

> My first goal will be to strip 2 items that XAR does not have, the  
> rpm lead and
> signature header. The lead will be discarded, the signature tags  
> will be
> appended to the metadata header (which is what currently happens now
> when a *.rpm package is read, see headerMergeLegacySigs in lib/ 
> package.c)
>
> At that point, XAR and RPM have a fundamentally equivalent archive  
> format
> consisting of contiguous metadata blob, plus a payload.
>

Achieved, at least the hackery.

rpm's appendix, the lead, is entirely vestigial, and likely headed  
for the
slop bucket permanently.

Nuking the signature header will need a bit more thought, but is highly
likely as well, XAR has its own concept(s) of signatures.

> This is likely to break everything on HEAD. You've been warned. ;-)
>

Nothing is broken. Yet.

Next will be a magic dispatcher to get rid of the pugly --nolead/-- 
nosigh/--newmagic
scaffholding. At the same time, I'll likely try compressed headers as  
well as XML
headers, as in rpm -qp --xml foo*.rpm, before starting the detailed  
mapping of
rpm <-> xar metadata.

Anyone for encrypted headers and files? It's formally the same  
problem as compression,
in fact is a simpler implementation, because symmetric encryption is  
length preserving.

73 de Jeff
Received on Tue Jul 31 19:19:53 2007
Driven by Jeff Johnson and the RPM project team.
Hosted by OpenPKG and Ralf S. Engelschall.
Powered by FreeBSD and OpenPKG.