[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Hijacking apt-cacher (Jonathan Oxer MIA?)

#include <hallo.h>

I do not get any answers from the maintainer of apt-cacher (Jonathan
Oxer <jon@debian.org>) without any obvious reason, he has been
responding few weeks ago.

Looking at the outstanding bug reports for apt-cacher, I decide to
hijack the package, where I rewrote the major parts (it's Debian native)
to solve structural problems. Jonathan has been promising fixes for
months/years now but I could not see any active development.

If anyone has contact with Jonathan, please tell him to contact me.
Otherwise I am going to upload the new package as 0.9 to unstable in
less than a week, declaring myself as the new maintainer. Snapshot of
the current changes file attached.


Format: 1.7
Date: Sun, 08 May 2005 23:07:38 +0200
Source: apt-cacher
Binary: apt-cacher
Architecture: source all
Distribution: experimental
Urgency: low
Maintainer: Jonathan Oxer <jon@debian.org>
Changed-By: Eduard Bloch <blade@debian.org>
 apt-cacher - caching system for Debian package and source files
Closes: 180544 203123 251468 251660 261273 267680 273776 274059 274975 277279 278070 278799 282599 294617 299404 305175 305956 307151
 apt-cacher ( experimental; urgency=low
   * NMU (most likely, no reaction from Jonathan)
   * new format, separates package contents and HTTP headers
     (closes: #274975). The new script apt-cacher-format-transition.pl converts
     the old cached files to the new version and moves the parts to the new
   * used syswrite/sysread where appropriate to minimise effects of Perl
     buffering in combination with Apache2 (avoids apt-get's long
     "waiting for headers" phase in most cases, still appears from time to
     time, but not soo often.
   * uses modification times of index files if configured, this should avoid
     desynchronisation of some files (closes: #180544). Used curl to get the
     HTTP head for that (wget was just too stupid with its timestamping
     abilities). By the way rewrote the fetcher code to use curl only, removing
     the wget depedency (closes: #277279)
   * rewrote large parts of unsafe code, worked around race conditions
     (closes:#251468), fixed some crap like inserting of status code into
     half-downloaded files (closes: #251660), really detached the fetcher
     thread from the reader when the file is initialy beeing downloaded, and
     made error code passing more reliable
   * removed another useless fork (thread-over-thread-over-thread, jeez...)
   * removed the CHLD handler that fscked up the return codes that I needed
     from close (became cruft anways since I dropped the unneccessary forking)
     It now also fails sanely on mirror failure conditions (closes:#203123)
   * allowing alternative URL scheme (with apt-cacher?/server/...) which does
     work with alternative http daemons and added alternative dependency on boa
     and httpd-cgi (closes: #282599, #273776)
   * applied patch from Peter Denison <bug-reports@marshadder.uklinux.net> for
     more flexible names of index files (closes: #267680)
   * IPv6 & filtering patch by Darren Salt (closes: #294617, #278070)
   * added my patch to do basic URL filtering (closes: #307151)
   * README.Debian update to the new stuff, removed cruft in debian/debian-old
   * rewrote the import script, made it work more efficient and work around the
     epoch numbers in the file names from apt's cache (closes: #278799)
   * rewrote and simplified the cleanup script (closes: #299404), also added
     support for source files and bzip2 compression (closes: #261273, #305956).
     Also made it refresh the index files rather then relying on possibly
     outdated data (or missing data because of tiffani/apt-dupdate usage) and
     really lock them while reading to not kill the cached data because the
     file is beeing downloaded just while the cleanup process runs
   * changed install.pl to copy the ownership of new files/directories and only
     doing so when they are new, rather than resetting them to www-data, and on
     every package upgrade
   * added my apt-precache.pl script for people that may need this toy
     (closes: #305175). It still needs some refinement to control the
     expiration of the "subscriptions".
   * added hooks for checksumming of forwarded packages
   * new feature: checksumming of data (downloaded and uploaded). Optionaly,
     see README.Debian for instructions to enable it (closes: #274059)
 f87111c2331a9bb05009312778433eb4 306 net optional apt-cacher_0.8.6.1+0.9beta1.dsc
 1a6d4d29351eb607cdf98ca13f06feaf 48627 net optional apt-cacher_0.8.6.1+0.9beta1.tar.gz
 3d4e6b1de9eb2923d2ca045e2b74ffcf 37048 net optional apt-cacher_0.8.6.1+0.9beta1_all.deb

Immerhin meint die Filmförderungsanstalt, im Jahr 2002 seien 59
Millionen CD-Rohlinge von 5,9 Millionen Nutzern mit Filmen bespielt
worden, im Durchschnitt also zwölf Rohlinge pro Anwender.
                -- <http://www.heise.de/newsticker/data/see-08.04.03-000/>

Attachment: signature.asc
Description: Digital signature

Reply to: