- Refactor into a support method and two API points
- Use std::transform for doing-things-to-a-list
- Add searchQmlFile that only takes a name, for
non-modules to use.
- Registration of QML modules may need to be done
for more parts of Calamares. Move into the library,
out of the model.
- Register for QML when using the QML sidebar.
- This is utility code, so it can be in the QML "service"
from Calamares, rather than in the QmlViewStep itself.
That makes it usable for other QML bits as well.
- The name is just the module identifier, and now we
search for *m@i* and also *m* from that identifier,
the name becomes much less important -- and it
can be set from the config key *qmlFilename* as well.
- To avoid name-collisions in otherwise well-behaved
modules and configurations, make the QML settings
more specific:
search -> qmlSearch
filename -> qmlFilename
- Having a ProgressTreeModel that does nothing but
proxy to ViewManager methods is kind of useless.
- Move the relevant code from ProgressTreeModel to
ViewManager.
- Remove now-unused ProgressTreeModel.
- The model is a simple list, not a tree (it may have been in the
distant past).
- All the information needed comes from the ViewSteps held by the
ViewManager.
- The delegate and fake-step handling was never used.
- Introduce convenience methods getString(), getBool() to pick
out an entry from item definitions in YAML format.
- Apply coding style.
- Pick up the "expanded" property as well.
- Use normal translation framework. The EncryptWidget was the one place
not using the "usual" translation framework, but rolled its own.
- Emphasize that the checkbox-state (checked-ness) is the parameter,
not a state of the EncryptWidget.
- All other instances of UI classes from Designer use a pointer-to-UI,
not multiple inheritance.
- Convenience method for setting the pixmap in response to
changes in the passphrase
- Tighten up types: enum -> enum class
- Reduce the scope for int-confusion by using an enum-class for
the encryption state of the widget
- Include UI implementation header only in .cpp
- Apply coding style
- Update copyright
- the strange construction of Helper and treating it as a singleton
can be factored out into a separate singleton-handling instance()
function. The Helper should never be destroyed.
- when a single function does more logging, it generally marks
those as subsequent debug-messages (with Continuation, or SubEntry)
and we don't need to print funcinfo for those, since it was already
printed the first time.
- there's no string representation for a QVariantMap, so it
won't be converted; in *debug* output it looks like there's a
string there.
- off-by-one when diving into compound selectors, spotted by
test, now fixed.
In order to test some of the internals, split them into Binding.h.
This makes the interface visible for tests. The implementation
still lives in the same place.
While here, adjust the test to the changed **example** which
now lists an additional variable.
- When a Python module calls utils.debug(), there's no point
in logging the C++ funcinfo that passes the parameters on;
don't use cDebug() with its attendant magic.
- Warnings, errors, don't get funcinfo, but regular cDebug()
calls do. Other special-cases, like calling Logger::CDebug()
constructor explicitly, don't get funcinfo either.
FIXES#1328
- Allow logging any QList type (needs explicit call in usage).
- Add a DebugList inheriting from DebugListT to keep existing
code that logs QStringLists.
- For Calamares 3.3, consider using C++17 and class template deduction.
- This bug has been here since f233cac7a1,
where a check for isSet() (of the -D option) was dropped. So since then,
Calamares has always been running with full logging (-D8) on.
- The recently-added "easter egg" of showing the debug-button when
log-level is 8 (to allow debugging-in-production) trips over the
default-log-level of 8, so the debug-button is always visible.
So, minor bugs in the debugging-setup, combine to show a debug-button
when there shouldn't be one.
FIXES#1329
- The manpage for umount says that -R can only be used with
a mount point (e.g. /usr/local) and not a device name;
this makes sense because a device might be mounted in multiple
locations, but the mountpoint (and things mounted under it) lives
in the filesystem tree.
- Existing code tried to unmount -R the device, not the mount point,
and so always failed; leaving things mounted that shouldn't.
Unset GRUB_SAVEDEFAULT if / or /boot is in btrfs or f2fs partition. This avoids the error "sparse file not allowed" at boot time. Btrfs and f2fs do not support saving default entry in grub.
- because mount() returns an exit code, and 0 is "success",
the if (!code) was backwards: when mounting succeeded, the
TemporaryMount object thought it failed.
- This leads to temp-mounts being left *all over* the place
from os-prober and fstab-handling.
- See editorial in the code-comment. Still need to test that
chroot(8) doesn't need a full path, otherwise this will
go to /usr/bin/env udevadm to force lookup (redundantly
if not in a chroot)
- !failed() also means "didn't write the file because it already
exists", which is sometimes acceptable -- but not here.
Use the more-strict bool() conversion, which is only when
the file was actually written.
- Unconditionally **not** overwriting the target file isn't an option:
writing hostname, for instance, expects that to be done even
if `/etc/hostname` already exists on the target filesystem.
- Make tests more resilient: do them in a temp-dir, and clean it
up after successful tests. This was prompted by tests failing
because of things hanging around in /tmp.
- Follow original patch from Gabriel Craciunescu: just drop
the *bufsize* parameter and stick to binary reads.
Text mode was associated in my testing with multiple hangs,
which didn't show up during binary-reads.
- By the time the GS is actually written, new (for partition state)
is always false. So "new" is the wrong thing to track. It should
have had a better name anyway,
- We store custom properties on the partition objects to indicate
what happens to them; use those properties (instead of state,
as done originally), call it "claimed" to indicate that the partition
is part of this installation.
For now, only new (as in, formatted, created-by-us) partitions are
claimed.
- The effect here is that only "new" swap will be added to the system,
so in erase-disk installations, or manual partitioning.
- Install-alongside and replace will now **not** claim the swap already
on the disk; I think we'll need another UI knob for that one.
FIXES#1316
- `createPartitionList()` is called for the summary widget (via
`prettyDescription()`), and from `exec()`. Only the latter
actually *writes* to Global Storage, so it's misleading to
think that the pretty-printed version ends up in GS.
- This makes the "new" key useless, since by the time `exec()` is called
the partitoons are no longer new.
- These tests don't actually test anything in this specific module,
they do test CalamaresUtils::System.
- Wrangling System and JobQueue and GlobalStorage instances is fraught
- Allow TranslatedString to get a context parameter; if it has
one, it will try to use the regular tr()-infrastructure
**as fallback** for the translations from the config file itself.
- This makes it possible to offer -- and translate -- some "standard"
phrases in the module, while allowing the config file the knob
to change strings. Using one of the standard strings gets translations
for "free", while introducing something entirely new means sourcing
translations for it as well.
- The model always has two columns, and the column names are always
the same. We don't need to specially set headers for that.
- Use QCoreApplication::translation() to re-use the existing
translations and avoid creating "new" strings (in a new context).
- Now that multiple netinstall pages may be supported, it's annoying
that they all have the same name. Copy the approach from other
modules (e.g. notesQML) of having the sidebar and other labels
configured in the config file.
- Since operations are added each time you leave this page,
the existing operations (from a previous visit) need to be
cleaned up. With the old setup of only **one** possible
set of operations, this wasn't a problem. Now, merging
in operations is necessary. Implement that by looking for
the *source* property in an operation.
FIXES#1303
- Different libraries should have different EXPORTs, so that
you can IMPORT one while building the other. Reported (and
kindly explained) by Kevin Kofler.
- Stick to one header file, though.
While here, update copyright on file.
- Having the widget do creation ties the step heavily to that UI;
start moving towards a state where we have a Config object (not
here yet; it still queries the UI part) that moves data around
between UI and ViewStep.
- This makes linking easier,
- Adds the right includes (needed on FreeBSD),
- Lets us drop silly GUI setting for non-GUI tests (I think this was
a side-effect of compiling on FreeBSD, where UI would pull in
/usr/local/include).
- Let's just have one header definining export- and visibility-
macros for Calamares. They are still selected based on the
export flags (*_PRO), just defined in one header instead of two.
- The scattering of DLL export macro's is kind of useless;
there are several headers, and then the export macro isn't
even applied consistently. Just drop the one for UI exports,
which was only used in libcalamaresui.
- If the test failed, you'd get a cryptic message like
FAIL! : NetworkTests::testPing() 'r' returned FALSE. ()
So rename the variable so the failure mode is more obvious.
(Could have used QVERIFY2() instead, this is simpler)
- Use the createTargetFile() convenience functions to do the
actual work.
- This probably involves more copying around of buffers, since it's
creating one big QString and sending that off, rather than writing
little chunks to a file, but I feel this is worth the code simplification.
- Drops all the error checking for creation, though, because the API for
createTargetFile() lousy.
Introduces a "partitioning service" into libcalamares,
shuffles a bunch of things into it, tries to help out
with settling the system between partitioning actions.
- explicit use of user-visible names in EditExistingPartitionDialog
- consistent conversion of config-values to FS names (user-visible).
The GS value comes from the ViewStep, and should always match
something -- it's already converted to the canonical un-translated
so the type should be good.
Because getting the untranslated name of a FileSystem is something
that needs doing consistently, add some functions for that;
it makes it easier to spot places where that isn't done.
Probably doesn't compile, and needs extra documentation.
- The config context object should be set earlier, otherwise
QML code will try binding to a non-existent config already
- Document that QMLViewStep::setConfigurationMap() parent implementation
should be called **last**, at the end of the subclass implementation.
- Using Branding::ImageEntry, when ImageEntry is an enum class
defined *in* Branding, is superfluous, and it also confuses
moc; the enum type isn't recognized from QML.
- Take the Python wrapper for GlobalStorage out of the GlobalStorage.h
header and add it to PythonHelper instead, saving some work in
all the cases that only GS is interesting, not the Python bits.
- Physical memory can't be negative, so it is reported as
an unsigned long, but the bytes-to-MiB functions do accept
negative amounts. As long as no machine has more than 2**62
bytes of memory, we're good though.
- don't have a NOTIFY CONSTANT property
- the data is constant, so drop NOTIFY
- remove redundant signals
- remove setLabels() now it's only needed from one constructor
- Create a job and ask it to create dbus files -- either directly,
or as a symlink. Since the target chroot isn't viable, this will
fail but we can at least see that directories are created, etc.
- Since these tests all want a system object, and a GS
with a sensible setup, give them one with its own initTestCase().
This could have been done with one executable, running tests from
multiple classes, but there's not much overall benefit there.
- Used to ensure that the directories leading up to a given path
exist. Implementation is incomplete and broken for now.
- While here, avoid removing an empty pathname in removeTargetFile()
(the empty pathname indicates a broken configuration).
- add a Settings::init() to do actual work
- remove the same kind of code from CalamaresApplication
- make constructor of Settings private
- initialize settings before the application
- There's a multi-stage setup for debug-mode, where the application
object also knows that debugging is set. Remove it.
- Break debug mode (because now the settings don't get debug-mode set).
- Refactor so that parameter handing is only done if this Calamares
is the unique (first) Calamares.
- To support translation testing, without needing to recompile
Calamares, load files from the local directory when debugging,
or from /usr/share/calamares/lang/ in general.
- This allows updating translations and testing them with just
lrelease (a translation build tool) installed, without rebuilding
Calamares.
- This allows distro's to ship updated or modified translations without
rebuilding Calamares.
- Most of the code was error-checking, just replace the open-read
with a call to the service instead.
- It's not an error if /dev/urandom doesn't exist in the source system
(there may be other good random sources, and otherwise we have the
low-quality random fallback).
- the list is already filtered for UTF-8, so this is redundant
- this *incidentally* fixes the problem with Assamese and Asturian,
since Assamese (as_IN) was having its only entry removed,
after which it would match Asturian (ast_ES)
- these were empty, so the widgets were hidden in the details
dialog of the requirements check; which looks really strange
if the reason the check fails is because root is required,
and you can't see that in the details.
This commit is on a branch because it changes strings, and I want
to do a release Real Soon and not annoy the translators.
- instead of counting and needing to keep track of the predicate
applied while creating the widgets, push nullptrs to the widget
list instead reflecting "this entry did not satisfy the predicate
for widget creation".
- for the list, the code can be the same as for the dialog,
only the predicate is different.
- while here, implement retranslate() since there's no text on
the list widgets otherwise.
- Create the label once, and it's ok for it to respond to links
even if there's none in the code.
- Turn into a member variable in preparation for retranslation-refactor.
- lift it out of the loop that creates the widgets
- some lambda-wankery, but the compiler hammers this down to
simple loops and you can read the resulting code as
none_of [the list] isUnSatisfied
none_of [the list] isMandatoryAndUnSatisfied
- no point in having init() called immediately after the constructor,
if it only makes sense to have one call to init() ever to create
the widget.
- while here, give it the same kind of structure as the dialog,
holding on to a reference to the list.
This is an ugly hack, using Bill Auger's support for Job weights.
The unpackfs job is arbitrarily awarded a weight of 12. That makes it
(in a Netrunner install) use progress from 12% to 40% or so, overall,
as all the files are unpacked.
Also fixes bug reported by Kevin Kofler that unpackfs was only reporting
progress when it hit an exact multiple of 100 (instead of over 100).
SEE #1176
- this is currently just an alias for QVariantMap, which is
the type already in use.
- future plan is to tighten this up and have an actual
Descriptor class that carries only the information
actually needed for the module descriptor.
- Replace stringlist with a stronger-typed list of InstanceKey objects
- Move smashing-that-to-stringlist into consumers of the list
(just one, the debug window)
sidebar entry can be configured and translated
adding a more elaborate qml example
keeping this in dummyqml for now, another commit will follow with
continuation of dummyqml in a more aptly named module
- introduce NamedEnum lookup tables for interface and type
- drop "final" and "virtual" from methods that don't make
sense as virtual
- shuffle declaration order so the virtual API for modules
sits together
- Trying to get away from untyped strings with special meaning.
- The "split identifier" branch tried the same thing, but
was duplicating the existing InstanceKey.h work.
- need a configuration before we can start loading (to support
the variable search paths)
- refactor showing a failure in the spinner widget. On failure,
the spinner will never go away, so a message for the user is good.
- stop clang-format from messing up the table of names.
- start of a class to hold configuration information; this can
later be substituted into the WelcomeViewStep and filled from
setConfigurationMap()
In the example application:
- register the Config type
- test application to display the QML (this will be extended
with adding the locale model to it)
- sample QML that does nothing useful yet (will display the locale
model once it's there)
- all the TZ location information now lives in the Calamares
locale service and the TZ list
- replace the Location class that was local to the timezone
widget by the TZZone class
- chase a bunch of small API changes that this needs
- Split the actual loading of translations into classes
to encapsulate the loading logic,
- Build a collection of classes to do the different kinds
of translation loading,
- Build a generic function to load something and update a
static pointer to the translation.
This makes installTranslator() much easier to read, and encapsulates
the type-specific loading somewhere else. While here, add a timezone-
translations loader so that the split-out TZ translations also work.
- Hide the one file from lupdate by giving it a weird suffix
- Call lupdate a second time for the timezone translations
- While here, adjust so that the options precede the directories
they are supposed to affect
I don't want to give the translation teams 444 new strings all
at once (about 90% of which don't need translation).
- Used in only one place, move to .cpp
- Drop useless scaling all the images *are* that size already
- Add debugging check that the images match expected size
- search for a key and return a type-cast pointer to the result
- while here, simplify some other code
- the find() function could be done with std::find_if but doesn't
get any shorter or more elegant
- By using QList< CStringPair* > consistently, we can save
a bunch of model code at the cost of an occasional dynamic_cast;
it's fairly rare for there to be a need for the derived pointer.
- read the file and create the regions on-the-fly, then sort the
resulting list (instead of building a string list and then
building the regions afterwards)
- needs a qwidget to put the top-items (license name, button) in
- fixes issue where the gap between the button and the hrule would
change depending on what is expanded
- Move layouting code into the .ui file
- Reduce margins hugely -- atop the title block, around the
scroll area, etc -- so that more license is visible at once.
- split shared <h1> message off
- do some string-concatenation, but only of whole sentences
- shave off some vertical space by dropping the mainsubtext item
- In code, add the necessary bool
- document meaning in the config file
- actually expand the full text if the entry is local and set to expanding-
by-default. This implementation is a bit lazy since it just pretends
to click on the toggle button.
- While here, reduce scope for UB by initializing POD members
- The arrows Up, Down, Right are used on toolbuttons, but
in the context of this module, those are directions with
meaning; give them better names.
- Because of #1268, the meaning of up- and down- may be swapped;
I'm not sure of which look makes the most sense. This is prep-
work for easily swapping the looks by using the meaningful names
instead.
SEE #1268
- we loop over all the entries anyway, so calculate allLicensesOptional
along the way (debatable whether std::none_of is clearer)
- always un-check the accept-box when resetting entries.
- Toggling the checkbox could disable the next button
because only the checked-state was used, instead of
the next-is-enabled-if-everything-is-optional member variable.
FIXES#1271
- Move retranslation to a separate slot to allow it to be
formatted nicely.
- Use calculated m_allLicensesOptional in retranslation.
- Untangle determining if all licenses are optional; std::none_of
returns true on an empty list.
- this isn't really a Calamares thing to decide, and anyway centering
on the desktop is kind of weird in multi-monitor setups and the
DesktopWidget is deprecated as well.
- Scenario: *keepDistribution* is true, and the existing file contains
a GRUB_DISTRIBUTION line **followed** by a commented-out GRUB_DISTRIBUTION
line.
- In that case, the commented-out line would change the flag back to
False, and we'd end up writing a second GRUB_DISTRIBUTION line at the end.
Prevent that: the flag can only go to "True" and then stays there.
Editorial: If your grub configuration would have tripped this up, then
you're doing something wrong. Clean up the configuration file first.
- If we update the line, then GRUB_DISTRIBUTION has been set
- If we don't update the line (e.g. because of *keepDistribution*)
then a comment doesn't count as "have seen that line".
This means that if we get to the end of the file, with only commented-
out GRUB_DISTRIBUTION lines, and *keepDistribution* is set, then we'll
still write a distribution line -- because otherwise it's not set at all.
- Previous fix would erase the distribution information (using an
empty string to flag 'preserve existing GRUB_DISTRIBUTION lines'),
but that is fragile. A distro might set that, and yet **not**
set a GRUB_DISTRIBUTION line, in which case it would end up with
a setup without any GRUB_DISTRIBUTION set.
- When a GRUB_DISTRIBUTION line is found, **then** check if it should
update the line or not. This way, we have a suitable distribution
to write if no GRUB_DISTRIBUTION is found at all.
- move the explicit checking for non-empty into a specific
(normal) password check
- leave only the-two-fields-are-equal outside of the password-
requirements framework
- having non-empty is the same as minLength 1, but gives a different
error message
- the two explicit checks are the ones that handle *two*
strings as special cases; all the other checks from
the password-requirements system only handle the one string.
- the explanations under and around the boxes is noisy,
hard to size correctly (viz. issue #1202)
- use tooltips in almost-all fields instead
- add placeholder text to be more suggestive
- since the wording of the checkbox itself (and the functionality)
is to enforce strong passwords, need to switch out some
logic and fix the wording of the documentation.
- The "convenience" method was no longer convenient, since
we now place strings on the buttons by default.
- While here, **name** the buttons so they can be themed.
- if the welcome module wasn't loaded (or loading otherwise failed)
then no text was set, leading to confusing screens with
buttons with icons but no label.
- If a module exists, and has unmet dependencies, then
that is only a problem if the module itself is *used*.
Merely existing is ok.
This triggers on FreeBSD, where partition isn't built, but
bootloader depends on partition -- so you can never start
Calamares on FreeBSD, because bootloader depends on something
non-existent.
Relax the check: just warn, and only fail if a non-existent
module is used (all those with unmet dependencies are considered
non-existent).
- Calamares scans **all** subdirs of the module-directory
for a module.desc and complains about those that don't have
a module.desc.
- For ./calamares -d runs from the build-directory, this
leads to a few complaints when some plugins have been
ignored (and so no module.desc is generated for them).
- Give the whole entry to file_copy, not just the
destination. This will allow file_copy to work
with local excludes.
- Pluck entry.destination out immediately, to keep
code changes minimal.
- Document the parameters.
- list_excludes() turns the extra mounts from global storage
into --exclude parameters for rsync; it doesn't do anything
with the destination parameter.
- while here rename to something more descriptive
- it's ok if item one creates directories where item two will write,
so don't check for existence of all directories on start-up.
Reported by ArcoLinux.
FIXES: #1252
This adds to the *machineid* module (which generates random UUIDs
for DBus and systemd) another key to configure generation of
a urandom pool in the target from the entropy in the host system.
- Improve documentation of the settings
- If sysconfigSetup is true, **only** setup sysconfig and ignore
the rest. This seems to be consistent with existing openSUSE-
derivative distro's, which set displaymanagers to something
nonsensical.
- the *mount* module inserts a rootMountPoint without trailing /
into global storage, so we can't assume that here. On the other
hand, the paths passed in to the Worker functions are absolute
paths -- adjust the tests to follow that.
- The code in Workers.cpp assumes that rootMountPoint ends in a /
so that it can have filenames appended easily; make the tests
fit that assumption, but still need to check that it is so in
production.
- refactor running the command into a helper function,
to deal with the regular if-command-failed-then-complain pattern.
- mark parameters as unused.
- move distinction about kind of DBus file up into the MachineIdJob
and remove the enum that marked it.
- Testing some of the functionality that's been added just now:
- copyfile fails, buggy implementation
- poolsize fails, buggy implementation
- removefile not tested
- read-urandom or copy-existing-file are implemented
- fairly chatty on failure
- needs tests (probably the implementation should be moved to
a separate file and unit-tested)
- keep the rootMountPoint and the path-with-random-data separate
instead of concatenating them at the beginning. Then we can
use the "clean" names also within the host system.
- this could be named isValid() instead, but basically the idea
is that this code makes sense:
JobResult r = do_thing();
if ( !r ) { /* Error happened! */ return r; }
/* Carry on .. */
- remove existing files for each kind of random-generation
that is enabled. There's a helper function for the case that
Cala is no longer setuid and needs help to remove those files
from the target (e.g. a setuid helper).
- Just use the existing rsync code, which can do both
files and directory trees.
- The existing code assumed we were always copying directories.
Now double-check beforehand.
- if a default DE is configured but the executable doesn't exist,
believe the .desktop file. Then use that, and warn if the
whole thing can not be found.
- for a DE entry which has a bad executable setting,
update the entry from the .desktop file using TryExec.
This assumes that the TryExec command is actually something
you might want to run.
- Moc generates Q_UNUSED(_a); which in turn (with clang) issues
a superfluous-semicolon warning. Existing code with automoc
uses utils/moc-warnings.h to turn off warnings that are issued
on moc code. Include it explicitly here because automoc isn't
applied.
- Sessions can be X11-sessions (living in xsessions) or Wayland-
(living in wayland-sessions). Look in both places.
- Refactor code a little to make it nicer to read.
- Drop the 1-argument QString constructor, it is suprising
- Drop the conversion to QString
- Add a toString() instead
- Drop tests for the removed API
- While here, apply code formatting to the tests
This is done to force consumers to update to strongly-typed
InstanceKeys.
- cover all the constructors
- Start with some tests that fail, showing bugs in the implementation
- Fix bug that "derp@derp" was creating a valid instance-key with
a bad module and id (need to use ::fromString() to get that
functionality).
- Extend tests with more bad cases.
- Refactor tests to simplify "this is bad" assertions.
- Things in libcalamares/ subdirectories are namespaced
according to that subdirectory (sometimes in namespace
Calamares, sometimes CalamaresUtils). Do that in modulesystem/ too.
- Do the async GeoIP checking in the async requirements-checking phase
- Do not return any requirements results -- we just need the async bit
- Drop the waiting widget, since it's not needed (done by the
requirements phase)
- If there is an item with id "" (empty), it is used as the
"no-package-selected" placeholder text.
- Existing code iterated over the abstract model and used the
name and description at the time the model was set -- but
by getting the name and description from the model, only
a single string was obtained instead of the full range
of translations.
- Therefore, when arriving on the page, the "no-package-selected"
information was displayed from the translation that was active
when the model was set.
Instead, extend the non-abstract model so we can find the no-package-
selected item and pass that explicitly to the page.
FIXES#1241
- Since the package chooser might be used more than once, or for
more specific items than "Packages", introduce a way to provide
specific strings for display.
- The only string needed is the ViewStep name, since the item with
id "" can be used for the no-selection item.
- My usual test environment has umask set to 022, but on one dev
box it is 002, leading to test failures (which show the test
was bad, not that the umask-setting code is bad)
- sometimes a slot is easier than a lambda. Introduce
a macro CALAMARES_RETRANSLATE_SLOT that calls a given
slot in an object on language change.
- extend the retranslator with support for calling slots:
- use Qt's signal/slot mechanism alongside the private
list of functions to call
- provide convenience for obtaining the retranslator of
an object.
- This helps give meaningful names to code chunks
- Gives clang-format something to work with
- Reduces indentation depth
I think this is a bit of a code-style opinion issue: big complicated
lambdas doing UI things just don't seem like a good idea.
- since we also need to *disable* the shortcuts, and should tell a
V1 slideshow that it no longer is running,
- use existing function to set the property to true / false depending.
- instead of changeState( true ) or changeStage( false ), use
meaningful enum names so that the code at the call site
becomes readable; make the boolean part internal to the
state-changing method.
- hangs unpredictably during testing
- replace with the Calamares process-invocation runCommand(), which is also
synchronous but doesn't hang (or, hasn't, in testing so far)
- The compile failure came from bad #include paths, so restoring
this interface declaration wasn't a fix.
- Reported to cause runtime failures on both KaOS and Manjaro.
If we don't have/need an image for the rootfs, we might want to
configure the `/` directory as a source for unpackfs. Unfortunately,
this raises an error:
- unpackfs first creates a temporary directory
- it then creates a subdirectory for each source, using the source
path's basename
- when the source is `/`, the basename is an empty string, therefore
the module tries to create an already existing directory
In order to prevent this error, we use the `os.makedirs` function with
parameter `exist_ok=True` instead of `os.mkdir`.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
- Synchronous download of a given URL; not something to
do from the GUI thread.
- Use it from the GeoIP service, which downloads in a
separate thread to do GeoIP lookups.
- Drop now-unused headers.
- Adjust tests for GeoIP to use network service
- The networking service is intended to wrap up use of
QNetworkAccessManager and others for consumption within
Calamares, and to provide some convenience functions
for internet access.
- Medium term, it may also monitor network access, so that
we can respond to changes in network availability during
installation.
Currently very minimal and undocumented.
- AppData and AppStream can be disabled independently of finding
their requirements (possibly useful if you want to ignore
AppStream even when it's installed in your build environment).
- Add a little top-level documentation about WITH_
- These don't have to be static methods of PackageItem, a free
function is more convenient.
- Since it's not API of PackageItem anymore, need to
- update tests not to use API
- do API-not-available warnings in consumers
- The smallest size image of the default (or, if there is no
default, the first) screenshot is used.
- Remote URLs are not supported by QPixmap, so most will not
load anyway.
- Use *appstream* as key in one of the items for the package-
chooser to load data from the AppStream cache in the system.
- Usable for some applications; for DE-selection not so much.
- Currently unimplemented.
- Put the implementation entirely in a separate file, keep the
not-supported one in PackageModel.cpp (but only in an #ifdef).
- Makes the various optional-data-sources more similar.
- Simplify the iteration by first determining which partitions
are mountable (at all).
- This guards against the very rare case that a partition
does not have a mountPoint at all (the if guarded against that)
where the lambda passed to sort() would get a KeyError.
Instead of having a special case for extra mounts to be processed right
after the rootfs, a better approach is to add them to the partitions
list, and then sort the list by mount point.
This way, we make sure every partition is mounted right when it is
needed: `/` is obviously mounted first, `/run` is mounted before
`/run/udev`, and so on.
The overall process is therefore more generic and should suit all
use-cases.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
- the "Select language" tooltip was applied to the form, so it would
show up inappropriately all over the place
- the buttons didn't have useful tooltips.
- having show*Url and donateUrl seems inconsistent, although
the show*Url settings were originally boolean-only.
- add "show" to the Donate button setting, to make them
all consistent (putting a boolean there will generate a
warning and hide the button, that's all).
- the generic (enum-based) setupButton() can handle all four
of the buttons, so setupLinks() can go away. Only the
(re)translation of the text on the button needs to be
done, so move that to the main RETRANSLATE.
- Handle buttons and their URL-opening in a more
general way with an enum; drop existing three-boot
method and special setupDonateButton()
- Doesn't compile because consumers haven't changed.
Currently, the `bytesToSectors()` function rounds a partition size to the
nearest MiB unit, which may lead to inconsistencies when a partition
is expected to only be a few KiB's.
This patch changes the behaviour of `bytesToSectors()` so that it aligns
on sector size, without rounding the partition size to a multiple of
1MiB.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
Currently, all size units are expressed as KiB, MiB or GiB (resp. 2^10,
2^20 or 2^30).
In order to maximize compatibility and consistent results with other
partitioning tools, this commit adds support for sizes expressed as KB,
MB or GB (resp. 10^3, 10^6 or 10^9).
This change won't affect existing users, it simply adds a new option
that wasn't previously handled.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
- this is not entirely straightfoward, since we need
different constructor arguments for the objects
Calamares creates (no QVariantList& args, in particular).
Implement our own registerPlugin() and createInstance()
for that.
- work around a bug in K_PLUGIN_FACTORY_DECLARATION_WITH_BASEFACTORY
As the config files integer are now of type `QVariant::LongLong` instead
of `QVariant::Int`, requirements relying on this type were not parsed
correctly.
This patch fixes this, and adds an option to the python conversion to
take into account `QVariant::LongLong` types.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
Currently, a number of configuration parsing-related functions and
classes use only `int` type for dealing with integers. Should the user
need a bigger integer value, this would result in an erroneous value
being used (`0`), as the correct value would overflow the 32-bits type.
In order to prevent these overflow, this patch replaces `int` with
`qint64` in the following functions & classes :
* CalamaresUtils::yamlScalarToVariant()
* CalamaresUtils::getInteger
* NamedSuffix
* PartitionSize
This way, sizes or other integer values greater than 2^31 (for signed
types) can be used.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
When the rootfs partition is read-only, mount points for the other
partitions cannot be created, therefore they need to be created in a
tmpfs, already mounted somewhere in `/`.
However, the extra mounts are only mounted at the end, which causes an
error as no tmpfs is currently mounted.
This patch makes sure all extra mounts are mounted right after the `/`
partition, allowing the use of a read-only rootfs.
Signed-off-by: Arnaud Ferraris <arnaud.ferraris@collabora.com>
This variable is declared in `if m:`. Of course if this codepath doesn't
run, the variable is not declared an Python doesn't like this kind of
surprise...
Signed-off-by: Arnaud Rebillout <arnaud.rebillout@collabora.com>
- If KPMcore is found -- it requires some other KDE Frameworks but
at least in pre-4.0 versions doesn't check very well for them --
then missing its dependencies is no cause for CMake failure.
Instead, log it nicely and suppress the module.
- stop compiler warnings with some judicious casting;
that's what you get when a container indexed by int
stored those indexes as quintptr.
- apply coding style
- do static initialization more carefully
- float -> qreal (double) because that's what the Qt API expects,
to reduce type-conversion warnings
- apply current coding style
- calamares_automoc() sets AUTOMOC, but also adds some flags
to avoid compilation warnings from the generated MOC code.
- drop weird hard-coded include paths
- Although milliseconds::count() is long long, we pass it to
a Qt interface that only takes int; let's assume we have
only a 32-bit count, since a timeout of 4 billion milliseconds
is roughly 46 days, which we'll just call "no timeout".
- Drop the round-trip of forming a JSON document from a QVariant,
then parsing the document into JSON objects and building a
model out of that. View the Variant directly.
- This is a fairly specialized class, for use only in the
whole-application where it ties in with the module system.
Move it to the application directory and slim down the UI library.
- Include it from the new location.
- Add UIC to Calamares (the application) because there's now
a designer-based widget in it.
- KDE and GNOME selection images were drawn by me for the bogus
package model, and should not be used.
- Keep Calamares logo though, even if it doesn't make much sense
to use it in package selection.
- Keep the no-selection image since it might be used more often,
but it's not very good.
- The ID and Screenshot entries might be weird in AppData (in particular,
a remove URL) so put those back under the control of Calamares even
when using AppData as the source of descriptions.
- Document all the static inline methods that do the work
- Fill up a QVariantMap from <name>, <summary> and <description><p>
elements, and use that to initialize the PackageItem.
- Doing a manual read of the XML, since existing appdata libraries
don't seem to have a convenient entry for what I need.
- Expand tests to loading AppData (currently, they fail).
- Massage the implementation a bit, don't insert a meaningless
copy of the key as the untranslated message.
- Add isEmpty() to check for presence of the untranslated message.
- Document API.
- Update tests.
- Add a FALLTHRU macro to annotate fallthrough situations in both
Clang and GCC,
- Annotate intentional fallthroughs.
- Add missing break which meant that the selection mode was
always multiple-selection.
- The tests should be run in C locale, otherwise the plain get()
function uses the current locale, which will fail (e.g. running
LANG=nl ./libcalamareslocaletest returns the Dutch strings for
plain get, which isn't what we expect).
- sr@latin is still special.
