summaryrefslogtreecommitdiffstats
path: root/urpm
Commit message (Collapse)AuthorAgeFilesLines
* Workaround removing real life kernelsThierry Vignaud2023-04-131-0/+3
| | | | | | | | | | Rationale: testsuite didn't covered the "kmod(vboxdrv.ko)" provides. So we were skipping eg virtualbox-kernel-6.0.9-desktop-1.mga9 because it provided "kmod(vboxdrv.ko)[== 7.0.4]" which was needed because virtualbox-7.0.4-1.mga9.x86_64 requires "kmod(vboxdrv.ko)" We need a better fix as in check if a (non orphaned) pkg still provides the property instead.
* (_all_unrequested_orphans) handle mga9 kernelsThierry Vignaud2023-04-131-1/+11
| | | | | Now kernel names are unversionned and we've multiple packages with the same "kernel" name which make orphans handling to derail
* (_replace_kernel_by_its_provide) Adjust for new kernel namingThierry Vignaud2023-04-131-1/+4
|
* (_replace_kernel_by_its_provide) split it outThierry Vignaud2023-04-131-2/+19
| | | | Rationale: so that we can document kernel management
* Track kernels by their NVRAThierry Vignaud2023-04-131-7/+10
| | | | | | | No change for mdv-mga8, but it's needed for mga9+ where multiple kernel with the same short name "kernel-foobar" are installed with different versions instead of having different pkgs names such as "kernel-foobar-5.12-1-1.mga8" and "kernel-foobar-5.16.3-1-1.mga8"
* (_kernel_callback) Describe it moreThierry Vignaud2023-04-131-1/+4
|
* Document that only urpmq uses get_orphans()Thierry Vignaud2023-04-131-0/+2
|
* Force timestamping=off when using wget to avoid conflict with --force-clobberMartin Whitaker2023-01-281-0/+1
| | | | | As seen on discuss@ml.mageia.org, if the user has set timestamping=on in /etc/wgetrc, wget fails due to conflicting options.
* Fix curl options to allow it to follow HTTP redirections (mga#31339)Martin Whitaker2023-01-211-6/+1
| | | | | | | 'curl -h' no longer outputs the full list of options, so the test for whether '--location-trusted' is supported always failed. As that option was added to curl in 2003, we can reasonably assume it is available.
* Prefer wget to curl as the default downloader (mga#24362)Martin Whitaker2023-01-211-1/+1
|
* Do not warn about 14y+ old behavior change (mga#26448)Thierry Vignaud2022-12-101-5/+1
|
* Simplify: those are from the same scopeThierry Vignaud2022-12-101-3/+3
|
* library: fix download with curl 7.86+ when the URL has a query part (mga#31047)Martin Whitaker2022-11-051-2/+11
| | | | | curl 7.86 changed the -O option to remove any trailing query part from the output file name. So if we have a query part, use the -o option instead.
* (migrate_forward_rpmdb_db_if_needed) Kill useless argThierry Vignaud2021-10-312-2/+2
|
* Remove uneeded argThierry Vignaud2021-10-312-6/+6
|
* Convert rpmdb to sqlite also when not chrootedThierry Vignaud2021-10-312-2/+2
| | | | | | | This is needed for live upgrades, either using urpmi or mgaonline (mga#29364). Two changes are needed: - fix detecting rpm version - fix converting rpmdb when not in a chroot
* SimplifyThierry Vignaud2021-10-311-2/+2
|
* Add a FIXME noteThierry Vignaud2021-10-311-0/+1
|
* Podify moreThierry Vignaud2021-10-313-36/+162
|
* Fix docThierry Vignaud2021-10-311-1/+1
|
* Rework mirror protocol selection to favour encrypted protocolsDan Fandrich2021-09-041-3/+17
| | | | | | | | Previously, HTTP servers were preferred over everything, meaning that HTTPS servers were never used. Now, HTTPS servers have an edge over other protocols, but if another server is much closer geographically than others, it will still be used. If so, HTTP (and FTPS) will have an edge over FTP and RSYNC.
* Use https: when getting the mirrorlistDan Fandrich2021-08-311-3/+3
|
* Document older downgrade conversion _after_ installationThierry Vignaud2021-06-051-0/+18
|
* Convert from bdb backend to sqliteThierry Vignaud2021-06-053-0/+56
| | | | | | Rationale: - sqlite backend is supported since rpm-4.16 - bdb backend is deprecated in 4.16 & removed in 4.17 (really 4.16.90)
* Clarity: librpm_version -> libbdb_versionThierry Vignaud2021-06-051-4/+4
| | | | | The original name introduced in commit d8cf4a35e631ca6ca54c18d247a5283cbb9bfcc4 was misleading
* Add --ignorearch to urpmi.addmedia --distribPascal Terjan2020-11-232-1/+6
| | | | | | | | urpmi.addmedia --distrib --ignorearch allows to use --distrib to create a chroot for a different architecture. This is useful to build for armv5 or armv7 on the other one, and to create chroots with qemu used to execute foreign binaries.
* Make it work with no feature qw(indirect) (aka perl7)Thierry Vignaud2020-06-252-2/+2
|
* Stop truncating disk names when prompting for a CDThierry Vignaud2020-05-062-4/+2
| | | | | | | .. or for a CD/USB key See mga#26564 comment 9: https://bugs.mageia.org/show_bug.cgi?id=26564#c9
* prevent forking a useless shellThierry Vignaud2020-04-291-1/+4
| | | | as shown by previous commit
* silent rpm --version during installation/removalThierry Vignaud2020-04-291-2/+3
| | | | | | | | | | | | | | | | Calling "LC_ALL=C rpm --version" from urpm::install::install_logger() sometimes causes the following to be displayed when using --urpmi-root: "shell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directory" Typically, we run into: chdir("/") = 0 chroot("/home/tv/mga/pkgs/RPM/perl-URPM/mga7/") = 0 (...) clone() execve("/bin/sh", ["sh", "-c", "LC_ALL=C rpm --version"], ... getcwd("(unreachable)/", 4095) = 15 So just caches the rpm version prior to the first transaction and voila
* cpan_testers: fix a test with rpm < 4.13.0Thierry Vignaud2020-04-091-0/+2
| | | | | | | | It was failing for older versions of rpm since commit 32c4bc874e3f7fc99188bf045902b1018f38855a This fix the counting & adjust to missing pkg name when erasing with older rpms
* add support for --reinstallThierry Vignaud2020-03-052-1/+3
| | | | Needs perl-URPM-5.25 (it won't work with rpm < 4.12.x)
* bump copyrightThierry Vignaud2020-01-1725-26/+26
|
* Fix test for whether any 32-bit packages are installed.Martin Whitaker2019-10-211-1/+1
|
* When adding distrib media, enable 32-bit media if needed (mga#24438)Martin Whitaker2019-09-281-11/+23
|
* Ensure downloaded files are world-readable (mga#24636)Martin Whitaker2019-06-221-0/+7
| | | | | mgaapplet and urpm* commands run as a normal user need to be able to read the downloaded synthesis and info files.
* Ensure urpmi config and cache files are world-readable (mga#24636)Martin Whitaker2019-06-212-5/+26
| | | | | | | | | This is needed to allow mgaapplet and urpm* commands run as a normal user to work when run from a gdm-x-session, which sets the umask to 027. See also mga#9386 and mga#22262. We should perhaps take notice of the msec security level.
* On systems with 32-bit EFI, enable the Core 32bit media by default.Martin Whitaker2019-05-272-2/+23
|
* Don't enable 32-bit media by default on 64-bit systems (mga#24376)Martin Whitaker2019-05-251-2/+4
|
* fix _rpm_version() when rpm is translatedThierry Vignaud2018-09-191-1/+1
|
* document open_helper()/close_helper() signaturesThierry Vignaud2018-09-171-0/+14
|
* add support for verify callback when installingThierry Vignaud2018-09-141-0/+1
| | | | Resolves: mga#23531
* typo fix in docThierry Vignaud2017-10-041-1/+1
|
* use XML::LibXML::Parser's 'huge' optionThierry Vignaud2017-09-181-1/+1
| | | | | | | | | | this is needed b/c of eg: ruby-google-api-client-doc which filelist is 11mb. However this makes libxml2 to fail b/c XML_MAX_TEXT_LENGTH is 10mb: $ urpmf-fast iso-grk2.gml Entity: line 4014501: parser error : xmlSAX2Characters: huge text node sr/share/gems/doc/google-api-client-0.13.5/ri/Google/Apis/SqladminV1beta3/Import
* update perl-base first tooThierry Vignaud2017-09-181-1/+1
|
* simplify: use //= where appropirateThierry Vignaud2017-07-171-1/+1
|
* Update Mageia copyright statementsRémi Verschelde2017-04-2625-7/+42
|
* Make sure that urpmi.cfg is world-readable so that mgaapplet can read it ↵Frédéric Buclin2017-04-031-0/+4
| | | | (mga#9386)
* skip kernel-firmware-nonfree tooMartin Whitaker2016-11-121-1/+1
| | | | | | | | | ...when considering real kernels Thus preventing kernel-firmware-nonfree being falsely detected as an orphan package (mga#12361). It should be handled by generic orphan mechanism
* fix more Getopt::Long breakagesThierry Vignaud2016-10-151-1/+1
| | | | now testuite pass again after droping the "gnu_compat" option
generated by cgit v1.2.1 (git 2.21.0) at 2025-08-11 15:38:00 +0000