Age | Commit message (Collapse) | Author | |
---|---|---|---|
2019-12-10 | Add manpages for OVERLAYS. | Bryan Drewery | |
Also reword the usage since these are not PATHS but registered ports tree. | |||
2019-12-10 | testport: Don't change PWD when running portlint | Bryan Drewery | |
2019-12-04 | It shouldn't be needed to export OVERLAYS (untested). | Bryan Drewery | |
Exported variables are special for distributed. If it doesn't need to be exported then it's less conflict for me to resolve later. | |||
2019-12-04 | Merge remote-tracking branch 'origin/master' into overlay | Bryan Drewery | |
2019-10-11 | testport: Show log info on failure exit | Bryan Drewery | |
2019-09-17 | Initial support for overlays in poudriere | Baptiste Daroussin | |
Only regular bulks tested | |||
2019-07-24 | Slight cleanups | Bryan Drewery | |
2019-07-23 | Remove make.conf after make config is run | Kubilay Kocak | |
2019-07-16 | Narrow make.conf setup scope | Kubilay Kocak | |
2019-06-04 | testport.sh: Fix make.conf not being respected | Kubilay Kocak | |
Currently, testport (-c) does not honour make.confs such as can be set in poudriere.d/{*-}make.conf files, because the __MAKE_CONF variable [1] is set to /dev/null, introduced in a43fe7cb to fix #545. Accordingly, when using testport -c, when dialog4ports is run, it presents information that does not match what is actually used/built because the expected make environment(s) configurations are not used. For example, this issue was identified while testing the www/py-autobahn port, and make config (dialog4ports) was incorrectly displaying py36-autobahn and py36 specific OPTIONS, when testport was in fact building py27-autobahn and py27 specific OPTIONS (correctly) via -zpy27, which contained a DEFAULT_VERSIONS=python=2.7 override. The issue is not limited to package names and OPTIONS, but anything that needs to take effect or be modified in the config target. Conversely, the options command introduced a fix to "Fix make.conf not being respected in 2d9b712b, introducing a setup_makeconf() function in common.sh that constructs a make.conf based on global, tree, and set specific make.conf's, if they are set. This change utilises the setup_makeconf() function in common.sh, and applies it to testport.sh to correctly respect make.confs when -c is used. [1] man make.conf | |||
2019-05-16 | buildlog_start: Swap params to match buildlog_stop | Bryan Drewery | |
2019-02-27 | log_start: Stop using global PKGNAME | Bryan Drewery | |
2019-02-27 | build_port: Stop using global PKGBASE | Bryan Drewery | |
2019-02-27 | bset_job_status: Stop using global PKGNAME | Bryan Drewery | |
2019-02-27 | buildlog_start: Stop using global PKGNAME | Bryan Drewery | |
2019-02-26 | build_port: Stop using global PKGNAME | Bryan Drewery | |
2019-02-26 | testport: Fix ports-mgmt/pkg Latest link leaking into real package repository. | Bryan Drewery | |
Fixes #617 | |||
2018-03-20 | testport: Show log summary at the end | Bryan Drewery | |
2018-03-20 | testport -c: Don't pass FLAVOR as a make argument | Bryan Drewery | |
2018-03-15 | MUTABLE_BASE=schg mtree: Don't ignore changes to schg files with testport or ↵ | Bryan Drewery | |
bulk -t | |||
2018-03-15 | Remove references to INDEX - support removed in 0bef52955606412ac64aceddde6 | Bryan Drewery | |
2018-03-09 | testport: Trim away excess '/' as done for bulk and dependencies. | Bryan Drewery | |
Reported by: koobs | |||
2018-01-03 | Improve security/parsing for MAX_FILES_<PKGBASE> | Bryan Drewery | |
2018-01-03 | Fix MAX_FILES_<PKGBASE> with hyphens. | Bryan Drewery | |
Fixes #572 | |||
2018-01-02 | Allow MAX_FILES_<PKGBASE> to override MAX_FILES | Bryan Drewery | |
2018-01-02 | Track PKGBASE internally | Bryan Drewery | |
2017-12-12 | testport -n: Implement as dry-run. | Bryan Drewery | |
This removes the old compatibility for the renamed -n -> -P flag. Issue #561 | |||
2017-12-12 | Update testport.sh | Zsolt Udvari | |
2017-11-13 | testport: Avoid reading ports files outside jail for FLAVOR/DEPENDS_ARGS | Bryan Drewery | |
2017-11-13 | testport -c: Don't read host /etc/make.conf. | Bryan Drewery | |
Fixes #545 | |||
2017-11-10 | testport: Undocument -s after its removal. | Bryan Drewery | |
Reported by: olgeni | |||
2017-11-09 | No need for _ignored here | Bryan Drewery | |
2017-11-09 | testport: Support MOVED with destination FLAVOR. | Bryan Drewery | |
Issue #541 | |||
2017-10-31 | Remove bulk/testport -s. | Bryan Drewery | |
This was a precursor to resuming builds with -B. It is not safe to use in general since it skips all incremental build checks such as changed options, changed PKGNAME, updated version, changed dependencies, changed default FLAVOR. -S remains which only skips removing packages if their dependencies are missing. This assumes that PORTREVISION bumps are done properly. -S can be used for when you trust PORTREVISION bumps are proper or that the change made to updated ports are all ABI-safe for downstream consumers. | |||
2017-10-28 | Follow-up 154aef06526: Add flavor to more verbose/job messages | Bryan Drewery | |
2017-10-28 | Record flavor for port build results. | Bryan Drewery | |
This shows in several places: - bulk final output - HTML tables - Some job status messages | |||
2017-10-20 | testport: Disallow testing @all | Bryan Drewery | |
2017-09-30 | testport: Add some specific tests for broken port or missing PREFIX. | Bryan Drewery | |
Fixes #531 | |||
2017-09-28 | testport: Fix time elapsed not showing in HTML listing. | Bryan Drewery | |
Issue #452 Reported by: sbz | |||
2017-08-29 | testport: Properly use BUILD_PARALLEL_JOBS (not PREPARE) for dependency ↵ | Bryan Drewery | |
building. Reported by: antoine | |||
2017-07-29 | buildlog_start: Reduce execs by using port_var_fetch for -V lookups | Bryan Drewery | |
2017-07-06 | Add assertion and comment missed for b40169fc671. | Bryan Drewery | |
Issue #512 | |||
2017-07-05 | testport/bulk -i: Fix not installing main package in jail. | Bryan Drewery | |
This was partially a merge conflict in 7c3831b2e, originally committed in cf73c49f4 with handling of ports->packages and proper was_a_testport_run logic checks. This also was a problem with where the shash data was stored before and after the prepare_ports call in testport.sh. The deps_fetch_vars lookup is moved to prepare_ports() now to handle properly creating the .p dir and setting SHASH_VAR_PATH before looking up port vars. Fixes #512 | |||
2017-06-18 | Apply DEPENDS_ARGS to dependencies in one place | Bryan Drewery | |
2017-06-18 | testport: Fix application of DEPENDS_ARGS to dependencies. | Bryan Drewery | |
LISTPORTS needs to be fixed up properly. It would probably be better if this were all done by deps_fetch_vars() already. | |||
2017-06-18 | Add more assertions on ports tree features | Bryan Drewery | |
2017-06-16 | testport: Fix cleaning phase showing wrong pkgname for DEPENDS_ARGS usage | Bryan Drewery | |
2017-06-16 | Always try mapping slave ports rather than only matching on '*py3*' | Bryan Drewery | |
2017-06-16 | Use PORTS_FEATURES to optimize lookups/support for DEPENDS_ARGS and FLAVORS. | Bryan Drewery | |
This also reworks some of the previously added assertions that FLAVORS cannot be combined with DEPENDS_ARGS since we don't ever lookup both now. | |||
2017-06-15 | testport: Fix testing on py3 slave ports | Bryan Drewery | |
This also allows testing on virtual (non-existent) py3 ports like textproc/py3-hexdump, but it otherwise is required to be able to properly build and map ports like net/py3-netifaces. |