- gcc (up to at least version 10) is worse at recognizing that all
cases have been handled, so it complains about all the switches
that cover enum values.
- both clang and g++ support __builtin_unreachable(); (as Kevin
Kofler pointed out) so we don't need the macro to do different things;
- the compilers have gotten better at detecting unreachable code,
so instead of inserting macros or fiddly bits, just drop them
and the unreachable code they comment.
- reduce the difference between clang and g++ builds, factor
common flags out of the CMake-if
- drop special boost-warning-suppression, we do that differently
most of the time in the affected source files
- if the partition size is invalid, then warn about it but do
not print the (uninitialized) size of the partition.
- shuffle code to continue earlier, allowing the "good path"
code to be out-dented.
The special setup for nicely-named groups which have a single
hidden subgroup containing the actual packages, has the problem
that there is a non-empty subgroups item, but this results
in zero actual children: then the number of selected and partly
selected children is also zero in updateSelected() and therefore
the item ends up unselected.
Special-case this to avoid unnecessarily unselecting the item.
Reported by Vitor L.
The status for an empty login name is '' (empty), for ok -- this is
so that there is no complaint about it. But it's not ok to
continue with an empty name.
The status for an empty login name is '' (empty), for ok -- this is
so that there is no complaint about it. But it's not ok to
continue with an empty name.
- The EXPECT_FAIL value "Abort" stops the test (I wanted 'if this
unexpectedly passes, raise an error' -- should have read the
documentation more closely).
- Set the shell in the config object, not just in GS.
- add a status member so the different steps can show progress
as the user is created and configured. The progress values
are hard-coded guesses as to how much work is done for each step.
- while here, reduce the scope of the global storage variable
This is somewhat experimental and weird; the idea is that bool
arguments are a lot easier to understand if there are proper
names attached, rather than "true" and "false".
This improves the situation for jobs that do not provide
a status: their blank status does not overwrite the status
bar, and since (previous commit) the description or name
is used to start the job if the status is empty, at least
**something** is displayed.
SEE #1528
- os-proper may return an extra file after the device:
/dev/sda1:Ubuntu 19.10 (19.10):Ubuntu:linux
/dev/sdb1@/EFI/Microsoft/Boot/bootmgfw.efi:Windows Boot Manager:Windows:efi
- The API definitions are just the symbols to define; these
are variously added through add_definitions() (needs -D)
or target_add_definitions() (doesn't).
I think we had this (kind of) module a long time ago and it was
removed for over-complicating things; re-introduce one now that
KPMcore is used in 3 different places and all would benefit
from consistent API handling / defines.
- handle swapfiles when writing /etc/fstab in the target system
- special-case mountpoint
- since swapfiles are not a partition, take the setting out
of partitionChoices
- create the physical swapfile as well (there's no other place
where it would make sense)
This still won't help if there's one really huge file that takes
several seconds to write, but if there's a bunch of files together
that is less than a file_chunk_count but take more than a half-
second to write, update anyway
If there's thousands of files in a squashfs (e.g. 400000 like on
some ArcoLinux ISOs) then progress would be reported every
4000 files, which can take quite some time to write. Reduce
file_chunk_count to at most 500, so that progress is reported
more often even if that wouldn't lead to a visible change
in the percentage progress: instead we **do** get a change
in files-transferred numbers.
- The total weight is only needed by the UnpackOperation,
not by each entry.
- Use a chunk size of 107 so that the number-complete seems busy:
the whole larger-or-smaller chunk size doesn't really matter.
- The progress-report was missing the weight of the current
module, so would report way too low if weight > 1. This affects
ArcoLinux configurations where one entry is huge and one is a
single file, so weights 50 and 1 are appropriate.
When there are multiple entries, the overall weight of the
module is divided between the entries: currently each entry
takes an equal amount of space in the overall progress.
When there are multiple entries which take wildly different
amounts of time (e.g. a squash-fs and a single file) then
the progress overall looks weird: the squash-fs gets half
of this module's weight, and the single file does too.
With the new *weight* key for entries, that division can
be tweaked so that progress looks more "even".
- parameter instanceKey was left over from previous work that
special-cased the weight of Python modules.
- while here, consistently do `~T() override`
With 1 CPU, Calamares still spawns 9 threads or so: eventloop,
dbus loop, QML loop, ... many of those are invisible to the
application. Contention occurs on startup when the UI is constructed,
and we end up with the module manager creating widgets alongside,
or ahead of, the main window UI. This can result in deadlock:
- in CalamaresApplication::initViewSteps
- in QML imports
This is partly because the signal-slots connections get "deep":
from loadModules() we emit *modulesLoaded* which ends up showing
the main window in initViewSteps(). Avoid this with a QTimer:
drop back to the event loop and release whatever locks are held,
so the QML thread can get on with it already. Then the timer
goes off and the view steps are created.
- Give LVM jobs a dummy argument Device* so that they
fit the functionality of makeJob for partitioning.
For those jobs that already take an LVMDevice*, this should
be the self-same device, but that isn't checked.
* Use the minSize when the target storage is smaller than the sum of sizes
* Percentage-defined partitions should be computed after setting hard-defined ones
This fixes issues when 0 byte partitions were created when the disk is too small.
Also fixes an issue with percent-defined partitions being forced to be defined at the end of the disk.
- create dirs as needed (this will normally be done by
unsquash, but for tests with paths it needs to be done
by hand)
- log what file is being checked
- filePath() doesn't like the absolute paths we have
(they're absolute in the chroot, and existing code
just sticks rootMountPoint in front)
Document keyboard change for Turkish F layout, and document
the keyboard configuration value better, with alternate
path used in e.g. openSUSE
FIXES#1397
Some tests -- notably the keyboard module -- need to have the
QRC for the module loaded as well (e.g. because of data in the
QRC). Add a RESOURCES parameter to calamares_add_test()
like calamares_add_plugin() already has, to build the
resources into the test.
Keyboard test now passes, since it was missing the data for
lookups before.
- both changing the autologin and changing the user (login) name
affect global storage, and both may need to change the autologin
username; split it into a free function.
- the fullname change was bypassing the login in changing the
login name, **but** then it needs a back-workaround to keep
the "custom" setting off (when custom is off, auto-fill username
and hostname is active).
- after loading the config, fill GS already.
- when finalizing GS, get the autologin settings again.
- setup the visibility and initial checked-state of the reuse-user-
password-for-root near where it gets connected; do similar
for the require-strong-password
- squash the lambda slot into the regular slot: no sense in
connecting twice to the same signal with the same receiver.
- only connect config->ui once
- only connect at all if the setting is visible (e.g. when weak
passwords are allowed for the require-strong checkbox, or
when root's password will be written for the reuse-password)
- switch to QStringList as parameter, since consumers (that is,
the debug dialog, which is what this is for) are interested
just in the **names** of the jobs.
- to allow mutex locking in const methods, mark them mutable.
- there's no need for a macro that is going to be used once,
especially if there's only one place it can be called.
- expand it in place and remove it from the installed CMake
module
once completed, this can be a fully functional (offline) locale selection option
worldmap.png no longer needed/in use
working is the stackview of the region & zones models
Timezone text bar shows correct timezone
currentIndex see comments on lines 65 & 139, not working
update of timezone text bar can't be tested if working as long no index is connected (see lines 93 & 168)
Still, already committing, since it does more then old Offline.qml, which had no function for timezone
- point to main Calamares site in the 'part of' headers instead
of to github (this is the "this file is part of Calamares"
opening line for most files).
- remove boilerplate from all source files, CMake modules and completions,
this is the 3-paragraph summary of the GPL-3.0-or-later, which has
a meaning entirely covered by the SPDX tag.
In spite of there being considerable documentation sometimes in the
config file, we go with CC0 because we don't want the notion of
'derived work' of a config file.
The example `settings.conf` is also CC0. Add some docs to
it while we're at it.
- the translations generated from public-domain files are CC0-1.0
- the files derived from Unicode tables are close to CC0-1.0,
possibly except that there is a FileCopyrightText line
- CC0-1.0 for the uninteresting version-headers
- GPL-3.0-or-later for the services
- add SPDX identifiers to Calamares C++ libraries and application sources
- add SPDX identifiers to Calamares QML (panels and slideshow)
- the `qmldir` is a list of names of things in the directory,
so CC0-1.0 it as "uninteresting"
- QRC files are lists of names of things in the directory,
so CC0-1.0 them as well
Some Calamares source files incorporate material from
3rd parties (unlike the 3rdparty/ dir, which is basically-
unchanged 3rd party source). Tidy up the FileCopyrightText
lines for those sources.
This is not an exhaustive effort.
There's lots of (YAML) test data that is just trivial configurations
for modules. Since the configurations themselves are **also** CC0-1.0,
and the tests are less interesting, license them equally liberally.
The build instructions are not that interesting, it's a toss-up
between CC0 and BSD-2, but because other CMake bits are BSD-2-Clause,
apply that to more CMakeLists. The copyright date isn't all that
accurate, but these are just inconsequential files.
While here, tidy up and get rid of some useless intermediates.
The .ui files are all GPL-3.0-or-later style, but it's
slightly difficult to keep licensing information in them:
it's XML, so an XML comment might work, but there's no
guarantee that safe/load will preserve them.
Put the SPDX tags in the <author> tag, so that it's visible
in Qt Designer.
This was causing CI builds to fail, since WEBVIEW_WITH_WEBKIT
is defined only in the Config file, not on the command-line.
This crept in accidentally while trying to get rid of that
config file entirely.
Re-jig the module-weight calculations.
- modules can have a weight
- module instances can have a weight
- jobs, from the module, can have a weight
This is now configurable on a case-by-case basis, rather than having
C++ only as an option and a weird hack for unpackfs.
This is more a test-inspired hack than anything else: since signals
are delivered asynchronously, we can end up delivering progress
signals out-of-order, and then the signal spy lists them wrong:
progress goes backwards.
Insert a tiny delay between jobs to allow signals to be delivered
in-order.
- compute weights and accumulations beforehand
- mutex-lock structures so you can enqueue while running jobs
- simplify progress reporting calculations
- doesn't actually run any jobs
reflect changes from users/Config.cpp
corrected id missing capital
mirror UsersQmlViewStep.cpp/h with the users versions
connections are still not made
In advance of PR #1491, test loading and stringlist extraction.
- from code, extraction works "normally"
- for YAML data, the stringlist isn't actually a stringlist
- if the user password is reused (or not) then check the
status of the passwords against the new reuse-setting
- if the allow-weak-passwords setting is changed, then
check the status of passwords (both of them) against
the new weakness setting
As explained by Kevin Kofler and abucodonosor, the
implementer line can carry a bunch of different values,
but none of them are actually interesting. Simplify
the code.
- the way isPasswordAcceptable was being used was buggy, leading
to test failures (now fixed)
- don't expose the function, anyway: it's an implementation
detail for passwordStatus() which in itself is an implementation
detail for status notifications.
- avoid update loops by checking values before emitting *Changed()
- check validity of user and root passwords when asked
- if root isn't going to be written, or re-uses the user password,
defer to those status checks.
- The weight is the module (instance) weight, which can be
- the default weight of 1
- the weight specified for the module (in module.desc / the module
descriptor; this defaults to 1, above)
- the weight specified for the instance (in settings.conf)
The last of these "wins"; weights are constrained to 1..100
The weight isn't actually used in progress computation yet.
- a handful of modules had an unused *requires* key in module.desc;
this is probably from previous intentions around
prerequisites-testing. Since the settings were empty anyway,
they have been removed.
- [unpackfs] Compacted the way *requiredModules* list is written
- loads emergency, noconfig, requiredModules keys
- warns (and marks descriptor invalid) if there are unused / unknown
keys left over in the descriptor data.
- add fields -- all const, all bogus -- to the descriptor,
introduce a stub method to load the descriptor from
YAML data (e.g. read from module.desc)
- lighten the type-naming in Module a little, with usings
- In most cases, you **know** the table covers all the enum
values, and the extra parameter *ok* is just annoying.
Provide a convenience that doesn't distinguish empty
from empty-but-valid.
- move the enums
- expose the named-enum functions for them
- **start** replacing Descriptor with something stronger; this fails
zero tests so it obviously wasn't tested at all
This module allows the generation of the initramfs in Alpine Linux based
systems (excluding postmarketOS). Very bare bones, but then again it
doesn't need much. It uses the Alpine Linux tool "mkinitfs" to do the
job.
- setting the weight in *instances* should be different from letting
the default weight (of 1) stand; explicitly saying 1 should
carry some weight (ha!)
- any invalid instance key will cause a complaint
- "new" custom instances in sequence get a complaint, but
the instance description added to the list is valid
- there's no reason to ignore custom instances that are **not**
mentioned in the *instances* section: it may be useful to
name more that one even without distinct config files.
- no more weights in constructors; do that in fromSettings() only.
- simplify test to drop those constructors
- set config file also for "normal" descriptors; fix test
module builds, installs and runs, connections are not working yet.
UserQmlViewstep.cpp/h are from PR https://github.com/calamares/calamares/pull/1356
sections are commented out to make the module build, but help is needed to get those 2 files corrected.
config names used in usersq.qml are guessed from users/Config.cpp
debug window shows correct entries in GS, and under the module tab, usersq.conf is read
as should too. Running shows most config used in users.qml are not registered/wrong,
many entries like: qrc:/usersq.qml:228:13: Unable to assign [undefined] to bool
It is understood not all needed from the users module has moved to Config.cpp yet,
but doing the PR now, since it runs, doesn't crash cala and help is needed to further implement.
- expose, for testing purposes, the load-from-YAML-data part
alongside the public constructor that reads a YAML file
- add test for building the list of instances
- looks funny
- is hard to get clang-format to respect this; it's intended as an
access-modifier, but those are baked into the code rather than
being configurable.
- is probably rare enough that #ifdef is acceptable
- was getting multiple definitions of moc-related code due to automoc
combined with KDSAG having its own #include moc, comment-out the include.
- while here, simplify the CMake bits for building KDSAG
- was getting multiple definitions of moc-related code due to automoc
combined with KDSAG having its own #include moc, comment-out the include.
- while here, simplify the CMake bits for building KDSAG
- Requirement.cpp was there "just in case" the header grew
functions that need an implementation, but that seems
unlikely (the header is just a struct of POD).
- the scripts are BSD-2-clause,
- the generated files are CC0 (I'm not *100%* sure about the
derived file CountryData_p.cpp, which lists countries and
country codes -- it **is** extracted from CLDR data which
is not CC0)
- QStringList doesn't round-trip correctly; add a test to
demonstrate that.
- Fix existing test to **not** use QStringList, but QVariantList
(of strings), which is how other code would use it.
The above is **kind** of moot because nothing uses the YAML-save
function, but it might.
While here, fix another test: YAML-loading can load JSON just fine.
- add apidox to all the untranslatedFS() methods
- add the most-basic of untranslatedFS(), which works on a given
FileSystem::Type; this one can handle special cases where
Cala needs a different untranslated name than what KPMCore provides.
- for the purposes of Calamares's nearest-location selection algorithm
for timezone selection, introduce spot patches: alternate markers
on the map to indicate "things close to here belong in this timezone".
- hide the implementation detail in the find() methods.
- Cape Town is in South Africa, so one might expect it to get South
Africa's timezone -- which is Africa/Johannesburg -- but Windhoek
is closer, so it gets that.
- Port Elisabeth is similar: Maseru lies between it an Johannesburg,
so it gets the wrong timezone, too.
These both illustrate how the limited resolution of the map, together
with the "closest location" lookup, can give poor results. For most
of South Africa, the "wrong" timezone is closer than the right one.
- The TZ widget uses a different coordinate system (mapping lat and lon
to pixel locations, and then calculating Manhattan distance from
that), so needs a different distance function.
- Simplify code: there's just one "closest TZ" function.
- introduce a distance function and use that, rather than coding it
inside the find() function. This is prep-work for unifying the
find() calls, based on various coordinate systems.
- create directories for new tests ahead of the tests themselves;
this **can** still cause problems if a test is run standalone.
- if creating the grub-dir at runtime is necessary, be informative
if it fails.
- the default_dir was only stored in modify_grub_default() to
create the directory if needed; move that functionality to
the get_grub_config_paths() function (and drop the "s",
since it now returns just one).
Go over the locale module again:
- new models that avoid weird casts and inconvenient iteration
- shared timezone data
- simple sorting and filtering
- simplify the map / QML version
FIXES#1476FIXES#1426
- test insert, remove, emitted signals
- test loading and saving of YAML and JSON
This shows up a big bug in the YAML saving code (which was never
used, it seems, anyway)
- the loadJson behavior did too many notifications, and was likely to
deadlock; write directly to the map instead and emit only once.
- the loadYaml method did something very different from its
documentation or intent.
- refactor into some free functions (out of the lambda's for connecting)
- introduce new method to call from onLeave(), matching previous
widget behavior
- we can do GeoIP and GeoNames lookups, **or**
- use Calamares's internal GeoIP lookup and country / city hints.
The online version is much more accurate, but costs more lookups;
in these examples, set it all to "offline" and document what needs
to change (code edit) to use the online version.
It's probably a good beginner job to introduce a bool in localeq.qml
to switch the behaviors.
- Config has suitable strings for displaying TZ information.
Use them and automatic bindings. Don't update the strings manually.
- Suggest online or offline TZ lookups based on what the distro wants.
Edit the QML to pick online lookups (needs access to the geonames
service, though).
- Drop the variables that point at config and geoip: the Config
object has a currentLocation, which is filled in by both the
configuration and any GeoIP lookup -- it doesn't have city
or country information though.
- status is a longer phrase
- name is a short human-readable name
- code is the internal code
Code that writes its own "Timezone set to" messages can use
the name, rather than the status.
- when no location has been set at all, there's no sensible TZ
to report; just leave it blank. In *practice* you won't hit this
code from the Calamares UI before a location has been set, because
the Config object is instantiated and then immediately configured,
but from tests or unusual UIs it could be.
- needs some massaging because Config otherwise depends on
ModuleManager which is a UI class (for the Reasons),
but we already have a BUILD_AS_TEST define for that purpose.
- demonstrate a nullptr deref.
- The Config object now uses the re-done models and timezone data
- most of the properties of the locale Config are unchanged
- much less complication in extracting data from the zones model
It's convenient when e.g. QComboBox::currentData() gets the key
"automatically", and the default role for that method is UserRole,
so let the value of KeyRole overlap.
The (renamed) class TranslatableString keeps a key value
(e.g. New_York) and a human-readable version around; the
human-readable one is passed through QObject::tr() for translation
on-the-fly.
- The models are overly complicated: **overall** there is just one
list of timezones, and we need various views on that list.
Start over with an empty model of regions.
- there **is** another source of information about the CPU,
so in the test use that to cross-check what hostCPU() says.
NB: it's probably a good idea to fall back on the same file
in hostCPU() for better accuracy.
- get username, password etc. from the config object, not the page
- jobs now depend entirely on config
- handle logic of "what's the root password" in Config
- The configuration for writing the hostname (to /etc/hostname,
to /etc/hosts and possibly to systemd-hostname) is read-only,
because it comes from the config file and won't change after.
- add the "reuse user password for root" setting to Config,
make the UI page follow that setting.
- add the require-strong-password default and toggle settings to
Config; this is not well-checked yet.
On the widget / UI side, connect checkboxes only if they are
visible; refactor reuse-user-password-for-root settings.
- The language code "ie" is not recognized,
- "ia" is, and it seems to be the post-war variant of
Interlingue, so we may want to rename / relabel.
The testEsperanto test -- now split into scripts and
esperanto -- would have picked "ie" out of the list
because it does map to C locale.
The TODO said it was unused: it **is** used, but only in
a very limited scope. Drop it from jobs where it wasn't
useful (e.g. those that just return prettyName(), outside
of the partition module).
This builds some machinery so that we can create
a detector for member-functions (methods) named <whatever>.
Use the macro to build the machinery:
DECLARE_HAS_METHOD(myFunction)
then after that,
has_myFunction<T>
is either std::true_type or std::false_type
depending on whether T has a method myFunction.
- no need for the definition to be in public header, move to implementation
- while here, sort the members and private methods
- add a makeJob() to add jobs to the queue
- having a struct with an obtuse API for adding jobs-that-need-to-happen-
to-this-device is just not good for maintainability.
- break the build by making things private.
- The enum for install choice was copied into PartitionActions and
used in the Config object; its definition does not belong in the UI.
- Chase the renamings required.
- add an option to select what button should be selected when the
partitioning module is started; TODO: the actual functionality is
**not** implemented.
- drop the previously suggested name, which didn't get beyond the
comments-in-the-config-file stage (but which intended to do the
same things as this one)
- add option to schema already, even if it's not implemented.
See #1297
FIXUP conf
The encryption widget (passphrase for disk encryption) should show
ok / warning / error whenever the state changes; this avoids
it showing up first with **no** icon (it should show a warning
when both passphrases are empty).
Both the KPMCore and the ChoicePage -- asynchronously -- were connected
to the nextStatusChanged() signal. So if the core said next was true,
that could end up communicated to the ViewManager, enabling the *next*
button in the UI.
Changing to the *erase* page generally triggers a KPMCore reload,
which later emits a `hasRootMountPointChanged()` signal, once the
layout is applied and the disk gets a root mount point. So we'd
get a `true` from KPMCore, which -- because it was connected directly
to the signal to the VM -- would override any other considerations.
Hook up both signals to an intermediate slot that just recalculates
whether the next button should be enabled, based on the state
both of the Choice page and whatever else.
- drop groups from the viewstep
- note that the Config object should also be in charge of creating
Jobs (but then the de-tangling needs to be completed)
- add tests of default groups loading
Doesn't compile because QRegExpValidator is a gui thing.
Previously, we check for RCC support every single time CMake runs.
This is slightly wasteful, and it wasn't being done right anyway.
But it's moot because:
- Calamares supports back to Qt 5.9
- Qt 5.9's version of rcc (at least, 5.9.7) **does** support the
command-line argument `--format-version 1`
- Everything newer does too.
Simplify translations a little, too: just use autorcc rather than
building things by hand.
- The sources were in src/calamares but processed and generated
in libcalamares, which is weird at best.
- Generate an "extended" version header.
- Use the extended version in the logger and nowhere else.
- While here, minor coding style cleanups
The overall change here means that after running CMake, only
Logger.cpp needs to be rebuilt (if the extended version has
changed) and not a handful of other files that don't need the
full version number, but do happen to include CalamaresVersion.h
- Very rarely do we need the full-git-version of Calamares,
so split that into a separate header with a little trickery.
- In the "normal" version header, drop the full-git-version values.
- This is a half-step: the ViewStep shouldn't do job creation either,
eventually it needs to be the Config object, but this is better
than asking the widget (UI) to create some jobs.
- When updating login- or host-name, or the autologin setting,
set it in GS as well. This is a minor improvement over doing
it only when leaving the page.
- Since the Config object isn't complete, there are leftovers in
the widget, which has a fillGlobalStorage() for the not-jobs-related
bits previously in createJobs().
- since the configuration is in the UI parts, we need the widget still
to load the whole configuration (until the config object is complete).
Create the widget before doing configuration; this is wrong. But now
we don't hit nullptr derefs all over.
- make the HostName textbox just a view on the Config's HostName
- make the username and login textboxes view onto Config
- query the Config rather than the UI for job data
- delay construction of the Page (widget) until it's needed
- hand the Config object to the Page on construction
This is prep-work for putting the configuration information into the
Config object, rather than in the UI elements.
- most of the things in utils/ are in the CalamaresUtils namespace,
let Permissions follow suit. Chase the name change in the
*preservefiles* module.
- add an `apply()` function for doing the most basic of chmod.
Note that we don't use `QFile::setPermissions()` because the
**values** used are different (0755 for chmod is 0x755 in the
enum value passed to `setPermissions()`).
On the "Users" tab, the user can choose a username. It was possible to
use 'root' as username, which led to an installation error, because
'root' exists already.
Added a new check to the username validation.
Fixes#1462.
With PR calamares/calamares#1357 the label of the "Manual partitioning" option
was changed, which introduced several downsides:
* The label is shown for UEFI and for BIOS installations.
* The mountpoint of the ESP is and should be distro specific.
* The label always mentioned GPT, which is irrelevant.
* The label should explain, what the option does, and not, what
problems can occur under certain circumstances.
set index in i18n.qml to -1, old settings were just for reading from the bogus model
current model uses strings, so index fails to read from it. This fixes cala crashing on loading i18n.qml
- Expand the documentation, emphasize octal-vs-decimal
- east-const consistently in this file (most of Calamares is west-const)
- shuffle the is-valid bool to the end of the data members,
so sorting by size.
- already had methods for various kinds of broken-up data, but
not one for plain "region/zone" strings; having this makes
it easier for QML to report a zone.
- use the region/zone method from QML, so that clicking on the
world map updates the actual TZ in Config.
- get network status from the global Network object; document that
- get the strings describing the language and LC settings from
the config-object instead of roll-our-own
- use the model of supported locales from Config to populate listboxes
- connect selection of language or LC to the Config object
- remove stray and useless TODOs
- remove unnecessary empty overrides
- clean up includes
- drop all the code that is now in Config
Since the business logic (setting locations, maintaining GS, ...)
is all in the Config object, the ViewStep is remarkably simple:
hook up a UI to the Config, which in the case of QML is done
automatically.
- since the Page hooked up a model and changed the region-selection
**after** connecting to signals, it would reset the location
to Africa/Abijan (alphabetically the first timezone) during
construction. Don't do that.
- this doesn't do the lookup **yet**
- while here, refactor setConfigurationMap so it reads like a story,
with chunks bitten out into a handful of static inline void methods.
- read the *region* and *zone* settings; this duplicates what
the ViewStep does and is currently unused, but ..
- add new support for using the system's TZ (rather than
the fixed values from *region* and *zone*). This complements
GeoIP lookup.
This is the actual feature that started the long rewrite of
the Config object (so that all the business logic would be in
one place, usable for both widgets and QML).
FIXES#1381
- writing *localeConf* settings to GS can be done always when the
formats are set, rather than special-cased. The code
that handles the "special case" of no widget existing for the ViewStep
overlooks the other crashes that happen then.
- Since Config knows what jobs to create, just ask it rather than
keeping a copy.
- allow finer-grained control over whether-or-not to adjust the
timezone in the live system.
- handle some special cases at the point of loading-configuration.
- document the setting in locale.conf
- correct some documentation bugs
- adjust the YAML schema for locale.conf so it's legal YAML syntax
**and** validates the current file.
- since all locale changes need to be entered into GS anyway, this
is something the Config object can do because it is the source
of truth for locale settings.
- drop all the GS settings from the Page.
- remove the weirdly-structured prettyStatus and similar:
the Config object has human-readable status strings (three,
for location, language, and LC-formats) which can be
normal properties with signals.
- Implement prettyStatus in the view step by querying the Config.
- the language and LC settings migrate from page to config
- add API for explicitly setting language (which is then preserved
when clicking new locations)
- since Config knows what settings there are, it should create the
jobs to run later -- not the Page.
- this doesn't work yet, because the Config does **not** know what
the selected timezone is yet.
The Config object wasn't being used at all in the locale module;
reset it to empty and start using it in locale, so that
configuration functionality can be added to it as-needed,
and with the necessary refactoring built-in.
In 022045ae05 a regression was introduced: if no *slideshowAPI*
is specified in the branding file, Calamares refuses to start, with
a YAML failure.
Before the refactoring, we had `YAML::Node doc` and looked up
the *slideshowAPI* in it with `doc["slideshowAPI"]`. After the
refactoring, we had `const YAML::Node& doc`. The `const` makes
all the difference:
- subscripting a non-existent key in a mutable Node silently
returns a Null node (and possibly inserts the key);
- subscripting a non-existent key in a const Node returns an
invalid or undefined node.
Calling IsNull() or IsScalar() on a Null node works: the functions
return a bool. Calling them on an invalid node throws an exception.
So in the **const** case, this code can throws an exception that it
doesn't in the non-const case:
`doc[ "slideshowAPI" ].IsScalar()`
- Massage the code to check for validity before checking for scalar
- Add a `get()` that produces more useful exception types when
looking up an invalid key
- Use `get()` to lookup the slideshow node just once.
- previously, the first column (name) was sized to show the
names **that were visible at startup**, which fails when
there are long names hidden in groups that are not expanded
immediately.
- change the columns to resize according to the contents; this makes
the descriptions jump to the right as the name column gets wider.
FIXES#1448
- there are no consumers for checking-the-capacity-of-the-drive
This parameter was introduced in 3cd18fd285 as "preparatory work"
but never completed. The architecture of the PartitionCoreModule
makes it very difficult to get the necessary parameters to
the right place, and it would probably be better to put
a SortFilterProxyModel in front of a partitioning model anyway.
Since the display code can already filter on size, just drop this one.
- make the rectangles slightly larger
- align text to center of the rectangle
- make the rectangle fill out the column; without this, the
width would collapse back to 0 after a change in the model,
which would draw 0-width rectangles.
FIXES#1453
- continuations, for the console, no longer print the date + level,
which makes things easier to visually group and read.
- the file log is mostly unchanged, except it contains more spaces now.
- fix the schema so the schema is valid json-schema
- the schema doesn't actually validate the *operations* yet
- sort the named backends (needs a double-check that the
list covers all the ones we currently support)
SEE #1441
- Don't do in code what is already done in the designer (.ui) file
- setFrameStyle() is difficult because it mixes different enums
into an int, which causes the warning from clang.
- Add access to the panel-sides membe of the view manager, and
calculate which sides are populated by panels (if any).
- Pass the calculated panel-sides to the view manager before it
starts adding viewpages, so they get consistent margins.
- Give classes a virtual destructor that need them
- Remove spurious ;
- Refactor addJobs() because that doesn't need to be in a class
- Remove redundant intermediate base-classes
- Some variant helpers take a default parameter if the map does not
contains the given key or if the type mismatches. Make all helpers
behave the same.
- Initialize the attribute partAttributes to 0; it is a primitive type
and it is not initialized in some constructors.
Fixes commit c1b5426c6 ([partition] Add support for partition attributes).
- Move implementation of default constructor to cpp.
added missing components listed as ResponsiveBase, ListItemDelegate & ListViewTemplate
parts of which were on nitrux
keyboard.qml no longer uses buttons within ListView, can't work as buttons and have them visible
see https://doc.qt.io/qt-5/qml-qtquick-listview.html#footerPositioning-prop
set ListView as actually visible within a normal calamares window size
- If the 'no tracking' box is checked, then the way to uncheck
it is to tick some **other** box.
- It doesn't make sense to unselect 'none' and then have .. none
selected.
KDE neon does not do this kind of tracking -- although it was originally
requested by KDE neon, no server roll-out was done once the
privacy policy was thought out.
- give the on-some-checkbox-state-changed slots better names
- while here, refactor is-any-actual-tracking-option-checked
- improve other debug messages, to be a whole sentence
- QString to-integer members detect if an integer string begins with
"0x" (base 16) or "0", base 8; but QVariant members do not.
- QString: the C language convention is used is base is set to 0.
- Convert to QString and use its member toLongLong() and set base to 0
to detect integer strings begin with a prefix.
- These have **not** been fixed for validation, so the schema's themselves
will fail to load. This is a consequence of variations in JSON-Schema
representations through various drafts. Fixing the schemata is
fairly straightforward.
This gives us 19 new tests, all of which fail.
- Note that this is missing *languageIcon* so if that gets uncommented,
it will fail validation.
- While here decide that should be
right up front in object (mappings) declaration.
- The size of a 2GiB partition (in bytes) is larger than the largest
32-bit signed integer; we hit signed overflow while calculating
2^11 * 2^10 * 2^10 and the test fails.
- Switch the whole table of sizes to qint64 instead.
- For testing purposes only, introduce a _qi suffix for qint64.
FIXES#1430
- All the headers go to relevant subdirs, but we don't keep
libcalamares and libcalamaresui apart.
- While here, remove unused variable from libcalamares CMake
- link the library privately -- the public API uses QVariantMap
- install FindYAMLCPP just in case
- add yamlcpp explicitly in the few places that really need it
(e.g. netinstall testing the parsing of netinstall.yaml)
- Add the Calamares CMake-modules to the search path automatically
- Export to CalamaresTargets.cmake and use namespace Calamares::
- Document imported targets
- Find Qt, because the translations machinery will need macros from that
- The installed lib links to IMPORTED libraries from KF5, so we need
to find them (again) as well.
- Do all the status indication in one component, but vary
the top-level message based on whether the mandatory
requirements are satisfied.
- Vary color and icon based on each requirement's *mandatory* setting.
- Value is configurable (through the "selector" which is passed
to GeoIP lookups). This is convenient for tests so you can "fix"
the value that the lookup will return.
- The Radio's are replaced by CheckBoxes and some logic, so
that different tracking styles can be enabled independently.
None of the settings end up in the Config yet, though.
- Uses global storage to steer the jobs that are created, in case
the slideshow needs to be tweaked by percentages or whatever.
- While here, add some code docs and apply coding style.
- When loading QML V2, both loadQmlV2Complete() and changeSlideShowState()
lock the same mutex, introduced in e7f4479df1.
- Explicitly unlock when loading is done and we need to change the state
immediately.
- For testing purposes, it's useful to load a module externally
and then register it to the ModuleManager (this hands off ownership).
- Refactor overall module loading to use the exposed single-module method.
- QML files need to be searched in specific places; this was initialized
by Calamares, but not for the text application. Move initialization
into the library.
- fixes:
12:44:25 [6]: Python Error:
<class 'TypeError'>
'builtin_function_or_method' object is not subscriptable
File "/usr/lib/calamares/modules/rawfs/main.py", line 188, in run
item.copy(filesystems.index(item), len(filesystems))
File "/usr/lib/calamares/modules/rawfs/main.py", line 99, in copy
if libcalamares.job.configuration["bogus"]:
- The -s will run the slideshow with a bogus job-queue, allowing easier
testing of the slideshow. This is more convenient than having a Calamares
with an empty show and a bogus exec section.
- The -s option for running the slideshow / execution phase of
Calamares needs to create a bogus Module for the ExecutionViewStep.
- Previously, unless setDefaultFontSize() was called explicitly,
the default size would be 0, leading to unexpected and weird
displays (and a warning on stderr).
- If setDefaultFontSize() is not called, get a sensible size instead
(like defaultFontHeight() was already trying to do).
- When loading *view* modules, we always need a QApplication for GUI
bits, because the widget for a module is created is very early.
- If it's a view module, replace the application object with one
that supports GUIs; without the --ui flag, though, it will just
run the jobs.
- All the configuration lives in the Config object (or the
tracking objects that it exposes).
- Get data from the config object for the jobs; TODO: give the
jobs a less-clunky interface.
The UI isn't hooked up to the Config object yet, though.
- a single tracking type can be enabled for configuration in the
config file; each must have a policy URL. Class TrackingStyleConfig
is a base class for that kind of configuration.
- root_mount_point was used initially for logging c1a139995 (adding new
bootloader job options are to use grub for BIOS, gummiboot for efi set
extra mountpoint when efi is found)
- the trace was removed since 533031b3c ([bootloader] print() does not
log)
- The Python configuration tests sometimes need extra setup, so
do that through a CMakeTests.txt file in the test directory.
- Patch up existing tests:
- grubcfg needs /tmp/calamares/etc/default to exist
- rawfs won't work on FreeBSD because of differences in /proc
- drop the *discard* from filesystems-on-SSD in the standard example
configuration.
- keep the table **with** *discard* around for referece and explanation.
Remember that the example configurations are intended as **examples**,
to document available settings, and do not reflect a sensible
production configuration.
FIXES#1395
clean up obsolete lines in welcomeq.qml
add requirement section from welcome.conf to welcomeq.conf
data shows correctly in Recommended.qml, fails to show any in Requirements.qml if run without admin rights
This makes it possible to remove QML from Calamares, possibly yielding
a smaller, lighter installer; it takes with it the nice slideshow,
modern configurable navigation and the QML UIs built for various modules.
By default, WITH_QML is on and the "normal" feature set is retained.
- look for Qml modules only when WITH_QML is on (the default)
- look for Network, since that's pulled in only implicitly
- disable the QML Calamares models (modules/*q) if no QML is
enabled; longer-term plan is to merge the **pages** back to
the "upstream" modules, and have things be run-time switchable,
but that's not here yet. Also disable the notesqml module when
QML is off.
- reminder to make all the ABI-relevant WITH_* settings available as #defines
- move the compilation of KDSAG to the calamares executable, not the library
- when DBus activation is on, drop all of kdsingleapplicationguard
- It is the requirements model (checking) that reports progress, and now
the model is accessible (ask for it with requirementsModel(), make the
messages come from there.
- The architecture of letting someone build up a list of requirements
from data emitted by the ModuleManager is broken: if it gets loaded
later, it will miss data; passing around complicated objects is
no fun anyway. Get rid of it, on the way to "ModuleManager has
its own model of requirements".
- Give the ModuleManager a RequirementsModel -- that is the source
of truth about the module-requirements of the modules managed
by that particular ModuleManager.
- Let the RequirementsChecker operate on a given RequirementsModel.
- The requirements are collected by ModuleManager, checked
by an internal RequirementsChecker and changes to the
requirements state are all signalled from ModuleManager.
By connecting the requirements in the welcome modules' Config
only to their own configs -- and immediately checking them,
which is bad on its own -- we end up with a disconnect between
what the ModuleManager says about requirements, and what
the welcome modules report on.
Doesn't compile (but I need to get it off this machine)
- Prepare to implement a picture-based slideshow alongside QML
- Split QML loading into the slideshow component
This might be good prep-work for moving QML loading out of the QMLViewStep as well.
fully implemented:
* loading of a live map, ESRI based, zooming & dragging possible
* IP address is translated to map coordinates
* loading of the map centers to the obtained coordinates, with a marker set
* coordinates are translated to a timezone, label visible at bottom of the map
* mouse movement will show changing coordinates
* clicking on new location will center map there, marker moved too, timezone label adjusted
* hasInternet switch set to either load Map.qml or Offline.qml
not done:
* get hasInternet status
* fill the fine-tune 181n.qml with proper locale & language data
* connect the obtained timezone to globalstorage
comments are left in the various files for what needs attention/changes
Improve testing framework while adding tests to rawfs to double-
check that the conversion (for Python 3.3 compatibility) of
capture_output is correct.
- If a module has tests/#.global or tests/#.job, these are used
as arguments to a test-run of loadmodule (which reads them
and runs the module with that configuration).
- This makes the old python-loading test and test-runner entirely
obsolete, so remove them too.
- By default, try to use DBus service to keep Calamares unique
- The older implementation via KDSingleApplicationGuard is still
available, just not used by default.
- Calamares doesn't like to run multiple instances, since they would
interfere with each other (stealing disks from each other, for instance).
The single-application code tries to prevent that.
- For -d runs, for developers where presumably they know what they are
doing, the single-application restriction is annoying: especially if
you need two instances at once for some kind of visual comparison.
Drop the single-app requirement if -d is given.
- If the QML navigation panel sets a height, use that
- If it doesn't, use 48px (which *may* make sense, but like many
screen dimensions in Calamares doesn't take HiDPI into account)
- Give the demo QML an explicit height of 48
- This code has existed for a long time but never stored anything
to the Branding object, and the most literal slideshow (just some
images) was not implemented.
Region "Africa" zones # 52
14:25:19 [6]: .. Zone "Asmara" QPoint(445,183)
14:25:19 [6]: .. First zone found 2 "2.0"
14:25:19 [6]: .. Also in zone 3 "3.0"
- All failures were being reported as Timeout, which is confusing
when they are not. Introduce HttpError for the not-timeout
other kinds of errors.
- Add operator<< for RequestStatus for nicer error logging.
- The Config object can handle GeoIP loading on its own. Both
View steps that used this had a derpy view->setCountry() that
didn't really do anything with the view anymore.
- This forces the EFI firmware to boot the loader that was just
created, whatever the boot order set in the firmware setup (USB,
CD/DVD, HD...).
- It is safe to use the first Boot Entry listed in BootOrder as the
previous command creates the new entry and adds it to the first place
of the BootOrder.
- This is a good example of being overly clever in C++
- the whole API with an enum requesting a specific string is a bit weird,
although it makes sense from the 'might need more strings specified'
point of view.
- sometimes if you use external OEM modules you might have those lines already present
- by skipping them you won't have double lines when rerun the module in a later step
Merge in the documentation from Bill Auger and then implement
what it documents (this had been missing; productWallpaper was
an orphan setting).
FIXES#1380
- If we have a wallpaper, bung in an extra QWidget between the main
window and the panels (sidebar, nav and main) where we set a
stylesheet that displays the chosen image.
- things that can be done in the designer file should be there,
not weirdly repeated in code elsewhere
- drop the insertion of an extra spacer (why not include it in the
designer file?)
- shuffle all the connect() calls down to the end of the constructor
This commit adds the new configuration `efiSystemPartitionName` to the
file partition.conf.
This option sets the partition name to the EFI System Partition that is
created. If this option is unset, the partition is left unnamed.
- Use << Logger::NoQuote{} to turn off quoting **and** the space
- In practice, in Calamares we use this only around other processes'
output, where we want neither quotes nor spaces.
- the test checks that the default locale is C or en_US .. let's just
make it so instead of relying on the environment. This fixes tests
on my dev-laptop, which happens to be set to en_NL (with volapuk
date format).
- Although we long ago replaced the getPartitions implementation, the
test is still there, and on a machine with no /dev/sda (e.g. because
root is on nvme) the echo-awk-shell-pipeline can give an empty string;
this is turned into a QStringList{""} which has one element, while
the new version has 0 elements.
- Special-case the test that empty strings should be empty lists, rather
than 1-element lists with an empty element.
reads data from languagesModel correctly, debug added to see index changes
code cleanup
button layout improved
install info text added, better spacing of text
about button no longer commented out, 3 reasons:
- info provided by any about is standard, will be odd if it can't be found
- about.qml is a seperate file, can be completely adjusted
- not showing it takes away the options for new QML cala users to see what is possible with Loader, thus stopping new possible contributors
- This is a follow-up to d0c205c1cc6a2ae49935c92bfd52911b9a0d43f7;
I really don't know why static constexpr const elements that are
not referenced by address need to be defined separately.
- rename enlarge to ensureSize() and change the meaning from
"make this much bigger" to "make sure this is displayed",
which is easier on the caller to calculate.
- the navigation bar was set "too tall", leave it at the natural
layout height for this widget
- margins needed some massaging to give contents some more space
(contents has a margin, so it doesn't need more space above
the navigation bar)
- This isn't something that Calamares can acutally fix,
so the test will be disabled later. After all, if
Brazzaville and Kinshasa are close enough that on the
map they are the same pixel, we can't move the cities.
- Merge all the format documentation into netinstall.conf,
where the example is given in full as an embedded
*groups* entry.
- Get README.md to point to the example.
- Fix up headers in netinstall.yaml, pointing to the
embedded example in netinstall.conf.
Timezones updated for these countries found on pixel detection tool
Dublin is in 0 (-1 needs editing)
Gibraltar should be 1.0 (0 needs editing)
Guernsey and Jersey are in 0 (1.0 needs editing)
Lisbon is in 0 (-1 needs editing)
Vilnius should be in 2 (1 needs editing)
- QImage needs Qt5::Gui, so this isn't a guiless-test; it can use
the offscreen QPA, though.
- Check that the images are all the same size
- Debugging / check code removed from timezonewidget
- The idea is to check all the TZ images for consistency, like
TimeZoneWidget::setCurrentLocation() does when DEBUG_TIMEZONES is
on; a zone-pixel should be set in only **one** image.
The test so far is just a stub.
- Make all four DEBUG_ flags actual CMake options, rather than
stuffing some of them in the rather-peculiar _enable_debug_flags.
Each debug option turns on suitable compile flags in the module(s)
that are affected.
- When writing YAML, given a float **always** write
some decimal digits (e.g. "1.0" rather than "1")
so that the type of the written-out thing stays
float.
- Avoids test failure with the sample `welcome.conf`
file which reads 1.0 and would write out 1, which then
led to type differences.
- Handle qlonglong explicitly
- Add a fallbackfor things that convert to qulonglong, to
avoid these remaining integer types from hitting the
very end of the if-chain, and being written out as
the **string** "<typename>"
When one of these common names for the netinstall page is used,
it gets pulled out of the standard translations, so that it
doesn't have to be translated in the per-distro config file.
These labels are common enough that they make sense for
everyone to have lying around.
FIXES#1367
(I say "fixed" but of course it's going to depend on the translation
workflow to make these available)
- In production, cDebug() might not show up, so the log will not
contain the lines saying what program is being run;
- Errors should at least mention the program name, but "env" or
"chroot" is not useful, so pull that from *args*, which is
the command we actually want to run.
- an empty command isn't going to work (although it might successfully
run chroot or env in the target system, that's not useful)
- while here, move variable declarations closer to their use.
- Slice overall progress into chunks, with each chunk of equal size
(as long as we have no overall count information) and place
the progress of the current chunk into its own slice.
- The entry knows where it should be mounted, and can remember that
- mount_entry() didn't use self, so made no sense as a method
of the Operation class
probably due to dynamically loading items
regionModel now lists, zonesModel only lists one delegate, but
working on QML modules can now continue without crashing cala
This doesn't actually **work** though, the QML uses older Calamares-internal
APIs and uses a ResponsiveBase that we don't have. Merge it mostly for
the Config and model changes.
FIXES#1355
- Add some extra checks for validity of m_currentStep (an index)
- Start off with explicitly invalid index, and keep it so until
loading is complete; this prevents the situation where quit-at-end
gets triggered after loading the very first module.
- io.calamares.modules doesn't exist
- ResponsiveBase doesn't exist
The module is now non-functional, but at least it loads and renders
a list of regions and zones.
- name default / example QML conventionally
- copy QRC from keyboard -- the QML needs to be included in the QRC --
since we don't want to have the QML in the keyboard module.
- follow branding settings, taking the panel-sides into account
- drop fixed width and height for QML parts
- give panels a minimum appropriate-dimension if they don't have one
This allows, for instance, putting both Widget sidebar and QML navigation
on the left-hand side of the window.
- Introduce an enum for panel-side
- Expose this to QML -- I can imagine that QML panels need to know
which side of the Calamares window they're on.
- Refactor loading the setting into a method that handles both
flavor and side
- There's no real reason to force the sidebar left and nav at the bottom,
certainly with QML supporting more layouts and being more flexible,
so document a mechanism to place the sidebar and navigation along
"edges" of the Calamares window.
- With an empty list, the question is meaningless
- .. and we called this with an empty list while constructing the
ViewManager; if quit-at-end is true, then this would terminate
Calamares immediately because the list was at the end.
- put signals in conventional place
- remove const int& parameter, that can just be int
- drop oddly-guarded code (that leaks memory); if the index (row)
being passed in, it's probably best to just crash
- remove unused signal warningMessageChanged
- Now the back button should be done by clients as well
- Refactor in CalamaresWindow to avoid leaking local button pointers
to surrounding code.
- Add macro UPDATE_BUTTON_PROPERTY for convenience in ViewManager
(ugh, macro) to change a value and emit corresponding update signal.
- add properties for the next button (enabled, label, icon...)
- update those properties as normal
- connect to the properties in the UI implementation
using Loader, can be reused for other widgets conversion
leave onClick example with full path, commneted out
title text for About is hardcoded, discuss option to make this configurable in welcome.conf
background color hardcoded, tested to work well in dark themes too
- since we've got two blocks of code copy-pasted, which both
decide to call one or the other of two member functions based
on a flavor value, turn that into a templated function.
- passing member functions looks a bit weird, and calling them
is syntactically surprising, but it cuts down the code a lot.
- the checker only collects and calls requirements; it has no
UI component, and only manages data (and a thread to do the
checking). Move it out of the UI library.
- this function lives in Module -- and is the only thing typing
Module to the ViewSteps and JobTypes. Split it out into its
own funciton. Nothing else in Module needs to befriend the
ViewSteps, so we move the friend declaration around a bit
as well.
- while here, apply coding style.
This is prep-work for moving module to libcalamares.
- drop the current and completed roles, and expose only
the currentIndex. QML can use the QObject property on
the model, while QWidgets can call internally through
the model's data() function.
- we don't need to provide role names for this, so drop that bit.
- simplify the delegate code while here.
