135715 2006-06-05 22:57 0000 media-fonts/{font-misc-misc,encodings,font-misc-misc}-1.0.0 - upstream changed tarballs 2009-07-11 17:36:21 0000 1 1 1 Unclassified Gentoo Linux Ebuilds 1.0 x86 Linux RESOLVED FIXED https://bugs.freedesktop.org/show_bug.cgi?id=7124 P1 critical --- 1 b.ohnsorg@freenet.de x11@gentoo.org ifreecarve@gmail.com jakub@gentoo.org musty.elessar@gmail.com ruckc@yahoo.com b.ohnsorg@freenet.de 2006-06-05 22:57:23 0000 When emerging media-font/encodings emerge states, that file already exists and no download necessary and complains in the next line about "Cannot download media-font/encodings-1.0.0.tar.bz2". I've looked it up and can find it under /usr/portage/distfiles/, has proper rights and owner/group and when I remove it, the same thing happens again (download occurs first, finishes correctly and then emerge complains again "Cannot download...") I also had a look at equery depends media-font/encodings and got not a single referer, description's not the best, too. So what is this package at least for? This happens, when emerging xorg-x11 (modular) Portage : 2.1_rc4-r2 glibc : 2.3.6-r3 gcc : 3.3.6 kernel : 2.6.16-r3 gentoo-base: 1.6.14 jakub@gentoo.org 2006-06-05 23:03:10 0000 *** Bug 135716 has been marked as a duplicate of this bug. *** jakub@gentoo.org 2006-06-05 23:03:17 0000 *** Bug 135717 has been marked as a duplicate of this bug. *** jakub@gentoo.org 2006-06-05 23:06:22 0000 Sigh, they did it again... upstream-- :=( jakub@gentoo.org 2006-06-05 23:16:01 0000 *** Bug 135695 has been marked as a duplicate of this bug. *** ruckc@yahoo.com 2006-06-06 04:04:02 0000 other packages this applies too media-fonts/font-misc-misc and font-cursor-misc b.ohnsorg@freenet.de 2006-06-06 08:43:21 0000 I'd say this is a serious problem, when I look at the amount of duplicate bugs and a certain frequency, these bugs occur. Maybe there should be a big fat warning or a better testing scenario. It doesn't seem to common knowledge, that for example ftp works in text- and binary mode. One adjusts linefeeds and encodings, to a certain extent, and one transfers bits, not characters. I can work around it when using --resume and --skip-first, but sometimes it's not possible and there'll be no consistency at all. Won't do the portage-dev scripts this job, create all the necessary checksums by downloading the tarball once? Maybe this should be added somewhere, before committing an ebuild, test it automatically, if checksums match, --fetch followed by a md5sum-call compared to the portage-functions. Help necessary? I got some dependency-checker too, to control revdep-rebuild (find executables and libraries, compare them to installed ebuilds or list ebuilds and find necessary files, incl. some basic ldd-check) dberkholz@gentoo.org 2006-06-06 08:50:30 0000 Fixed, thanks! If you have an issue with the digest process, Onkobu, please file a separate bug. ruckc@yahoo.com 2006-06-06 10:36:19 0000 What was the resolution to fix this? Were the digests updated somehow? Were the files that we download replaced? dberkholz@gentoo.org 2006-06-06 10:41:45 0000 The digests were fixed. jakub@gentoo.org 2006-06-06 12:38:07 0000 *** Bug 135812 has been marked as a duplicate of this bug. *** jakub@gentoo.org 2006-06-07 03:02:12 0000 *** Bug 135878 has been marked as a duplicate of this bug. *** b.ohnsorg@freenet.de 2006-06-08 23:13:34 0000 Me 'ave no problem wit digest process. But I'd say it's a common problem and happens on and on again. So there should be a bit more improvement than fixing digests. Maybe all ebuilds must pass a fetch-only/checksum testcase, before becoming available to the masses or getting a '+' (stable). If it is a problem of the upload process (which may happen automatically) or if it's caused by the project owners (who change tarballs after releasing) this should be discussed (to keep problems away from gentoo's responsibility). Only an idea, nothing serious...