RFEs: 1. feed rpm --root patch upstream to mainline busybox 2. formally example/doco rpm2cpio.sh in /usr/share/doc/rpm-4.2/ 3. make busybox rpm md5sum aware and safer [00:03] <gmkurtzer_> I said it only because I saw it at http://www.busybox.net/screenshot.html [00:03] <jbj> really, time to putter off and look, easier than putting rpm on diet if already done. [00:15] <jbj> yah, rpm and rpm2cpio both in busybox, sufficient to get job done too. lots of metadata ignored, other problems, but basically workable. [00:19] <gmkurtzer_> notice the rpm in busybox does not support chroot [00:19] <gmkurtzer_> Where is that diet-rpmi when you need it! ;) [00:20] <orc_orc> gmkurtzer_: chroot command is in busybox -- wing it [00:21] <jbj> gmkkurtzer : you need/want --root? that shouldn't be more than 10 lines of code, not hard. [00:22] <orc_orc> jbj: can you consider upstreaming that to mainline busybox as an RFE, please? [00:28] <jbj> orc_orc: keep poking me, i'm hoping to do internal rework of parts of busybox in forth/lua as proof-of-concept bit slicer (like asn1) before putting into rpm. i can do the chroot thing in maybe 15 minutes somewhen, mebbe pick up filemode and mknod(2) from metadata too, that's probably worst hack, they trust the mode in the payload headers. [00:29] <gmkurtzer_> I just heard back from Eric (author of busybox), and he would really appreciate the root patch. I will look into, but I am out of my realm here. [00:29] <jbj> hmmm, i suspect md5 check on extracted files could be done pretty simply too. [00:29] <gmkurtzer_> Also, he mentioned to me how to get the support. [00:29] <gmkurtzer_> for rpm in busybox. [00:30] <orc_orc> gmkurtzer_: send an IRC scrape to him -- [00:30] <jbj> orc_orc: hit me with an RFE in bugzilla please, bugzilla never forgets ;-) [00:30] <orc_orc> gotit [00:30] <jbj> bugzilla.redhat.com
RFE is against busybox's rpm implementation.
The last version of busybox package (busybox-1.01-2) does not contain rpm command so these problems are not present there.
Ivana Of course busybox does have 'rpm' in it -- the option is not turned on in the build in the version for anaconda, as it is not needed there, but rpm is clearly in core busybox This bug (and its three part RFE) was filed at the request of Jeff Johnson, for some needs in the chroot build process. see his comment at 00:28 in the transcript Please look at the busybox sources and build model -- the need is to emit a patch for the upstream busybox maintainer, who has indicated he will take it, from the baseline RPM implementation, so taht the --root environment functions in busybox, as it does in rpm, native. .. reopening
Busybox contains 'rpm' but fc version does not choose it as a part of itself, so rpm problem does not affect fc or rhel. I'm sorry but I don't know which patch Jeff meant and Jeff Johnson does not work here so I can't ask him. I'm closing this bug.