Age | Commit message (Collapse) | Author |
|
Produce an error message for untest|vault with invalid package name, rather than a python backtrace.
|
|
Future work: make these regexes so we don't have to add them again in a
years time.
|
|
|
|
|
|
Drop Eric Blake as inactive
Add Achim Gratz
|
|
Also add equivalent config keys for cgit
|
|
|
|
|
|
Previously, we'd do some rearrangement of the release area, so the
obsoleted package would be moved under the obsoleting package, so it
didn't need to exist at all in the package list and we could safely
ignore it.
We don't need or want to bother with that now, so let obsolete packages
paths exist.
Future work: This needs some thought about how we want to work. Perhaps
prohibit uploads? Perhaps error if the package marked OBSOLETE isn't
actually obsoleted by something?
|
|
|
|
Also: isolate each individual deploy, by putting them in a subdirectory
of stagingdir named after the job id.
|
|
We want different treatment for files which are manually uploaded (into
homedir) and files which arrive from the build service deployer (into
stagingdir)
Broken manual uploads were left in place so they can be inspected and
manually fixed by someone.
Currently, broken build deploys will also hang around forever, but they
are expected to be "correct", and the initator of the build has no
access to the stagingdir to fix things. Instead make them get
automatically removed.
|
|
The data ordered by package name is the one we use the most often, so
make that the primary form.
Also use mtime_cache for the result of parsing cygwin-pkg-maint.
v2:
Fix mkgitolite and reports
Keep maintainers in a list, not a set, because we need "first named
maintainer".
Don't write permissions lines with an empty list of names, as that's
syntactically incorrect.
|
|
If post-stale removal package set validation fails, don't change the
packageset during process_relarea(). This makes it similar to
process_uploads, in that the packageset can't be changed to something
invalid.
(Scenario: vaulting is requested via 'vault' command, which triggers
process_relarea(), which applies any requested vaultings during it's
stale evaluation. If that fails to validate, the packageset is broken
and any subsequent uploads will fail until after another relarea scan
takes place)
Fixes: 0939d5bd86f4 ("Add 'calm-tool vault'")
|
|
|
|
|
|
|
|
Fixes: d77b94da7803 ("Propagate any error reading packages through mksetupini")
|
|
Drop a spurious argument to maintainers.add_packages(), left-over after
ad6b47c27b11 ("Relax trusted maintainer restrictions").
Also report an error if CYGNAME isn't set.
|
|
Don't add a custom keys to license_expression's database, if it's
already present.
Also drop unneeded 'XVIEW' license.
|
|
At the moment, calm.process_relarea() is expected to keep going,
although we might later fail due to package set validation problemss.
(In particular tests just assume that we ignore various problematic
packages, although we don't actually check that they are being rejected
with the epxected problem)
|
|
Rename orphanmaint -> trustedmaint
Also revise and relax logic so it's more consistent: trusted maintainers
can do these things via a shell, so don't stop doing them more easily
via calm.
Drop convulted "add trusted maintainers as maintainers of orphaned
packages so they can upload them", and just check directly against
trusted maintainer list to determine if an upload is permitted.
|
|
|
|
We removed all the !packages files some time ago.
maintainer-keys/add now takes care of ensuring the appropriate
directories exist when a maintainer is added.
combine identifies defunct maintainers, using a different criteria, but
similarly doesn't do any clean up.
|
|
In f3a2daab817a, we accidentally allowed packages to depend on a package
which has been removed, instead of just applying that to obsoletes. Fix
that so we check that requires: contains packages which actually exist!
Also clean up no longer required 'splitchar'
Fixes: f3a2daab817a ("Persistently record all package names")
|
|
Just as package name, only allow alphanumerics and '-._+'
Warn about other characters, just as we already warn about '-', and add
an exception for the one existing package which breaks these rules (with
a ~).
Also: '._+' should be allowed in R
|
|
|
|
|
|
This can remove packages which aren't installable any more, as they rely
on a provide which has had whatever provided it expired.
|
|
(since we're going to annotate python26 dependent packages which are
missing that, to facilitate their removal)
|
|
|
|
This probably isn't always wrong, but seems like it's something we don't
want or handle well at the moment (the solver will pick exactly one
replacement package to install, since that's all that needed to satisfy
the provides).
|
|
|
|
|
|
|
|
ruby-atk-debuginfo is obsoleted by ruby, but depends on
cygwin-debuginfo. The existing obsolete is the correct one.
okular-odp has a strange anomaly where it's category _obsolete, but the
oldest version depends on calligra-libs as a replacement, but more
recent versions depend on okular-calligra, which is correct.
|
|
self-destruct
|
|
|
|
|
|
Allow a package version to be explicitly marked as weakly retained in
override.hint.
Ideally, we'd like something that let's us make a transaction that
removes a given source package and all install packages generated from
it, but we're not there yet...
|
|
Allow an empty source for obsoleted catgets package, so libcatgets1 can
get properly identified as a deprecated soversion.
|
|
|
|
Drop special case to handle libturbojpeg as a solib package, since it
got expired.
|
|
'libflint' and 'libturbojpeg' packages were mistakenly first made as
solib packages without a soversion number. Explicitly match them so
they get treated as old solibs
|
|
flake-import-order"
This reverts commit 1d345508ee7ed441330da4a9810f21af216c6e58.
|
|
Rather than treating the sha512.sum file as (kind of) part of the
package, just treat it as a cache for sha512 computation of files in
the same directory.
Also use the mtime_cache decorator to avoid parsing sha512.sum files
which haven't changed.
|
|
|
|
|
|
|
|
|