This entry roughly documents how to get Etcher running on Wayland by
using the XWayland Server.
Fixes: https://github.com/resin-io/etcher/issues/509
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
This PR fixes an uncaught error being thrown when
`OSOpenExternalService.open()` is called with an undefined value.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, the logic the controls the drive scanner and populates the
drive model lives in the main application controller.
This is not optimal because:
- The drive scanner stops populating the drives model when the
application is not on the main screen.
- An event handler subscribes to the drive scanner every time the user
navigates back to the main screen.
This PR moves the drive scanner logic to a run block in the entry point
of the application.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Error handling is currently a mess. The knowledge to correctly report an
error both to the end user and to us is scattered in many places.
This PR introduces the following changes:
- Rename `AnalyticsService.logDebug()` to `AnalyticsService.logDebug()`
to clarify better the intention of the function.
- Move `$log` decorators from an `AnalyticsService` run block to
`AnalyticsService.logDebug()`.
- Implement `AnalyticsService.logException()`, whose duty is to log an
exception to TrackJS or any related service, and log it to DevTools.
- Implement `ErrorService.reportException()`, whose duty is to report an
exception to every interested party. This means logging the error to
TrackJS, displaying it DevTools and showing a nice alert to the user.
This function is based from `handleError()` from `MainController`.
- Move global `$exceptionHandler` error handler to the entry point of
the application, and make it simply call
`ErrorService.reportException()`.
- Replace every `handleError()` call in `MainController` with
`ErrorService.reportException()`.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
The problem can be reproduced with the following steps:
- Start Etcher.
- Plug a single drive, and let Etcher auto-select it.
- Navigate to the settings screen.
- Go back to the main screen.
The drive is auto-removed for a very small amount of time, until
auto-selection takes care of selecting it again. This behaviour causes
no harm, but its a bit annoying.
The problem was caused by a previous fix, which auto-removed the drive
if navigating back to the main screen after a flash finished, however
the fix no longer makes a lot of sense now that we allow a drive to be
selected even when no image has been chosen.
Change-Type: patch
Changelog-Entry: Prevent selected drive from getting auto-removed when navigating back to the main screen from another screen.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, we show "Finishing..." indenpendently on if we're waiting for
the flash to emit the "done" event, or we're waiting for the drive to be
unmounted.
As a way to simplify things, we move the hairy `ngShow` stack from the
main template to a nice function in the controller.
Change-Type: minor
Changelog-Entry: Show "Unmounting..." while unmounting a drive.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
There are some cases where AngularJS won't recognise that the available
drive list has changed, and incorrectly keeps asking the user to connect
a drive.
This problem was mainly witnessed in Windows, and could be reproduced
with the following steps:
- Start Etcher with no available drive.
- Select an image.
- Plug a drive.
Etcher keeps showing "Connect a drive" in the second step.
Change-Type: patch
Changelog-Entry: Fix new available drives not being recognised automatically in Windows.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
When using the `--robot` option, we write to `stdout`/`stderr` using
`WritableStream#write()`. This operation is asynchronous, however we're
not waiting for the data to be flushed before calling `process.exit()`
right after emitting the `done` event.
This causes a rance condition where sometimes `done` is never written to
the log file, and therefore the GUI remains waiting forever.
Change-Type: patch
Changelog-Entry: Fix application stuck at "Finishing".
See: https://github.com/resin-io/etcher/issues/613
See: https://github.com/resin-io/etcher/issues/609
See: https://github.com/resin-io/etcher/issues/573
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, if no graphical polkit authentication agent is found,
`sudo-prompt` will yield a "User did not grant permission." error, which
gives the impression the user cancelled the dialog (or entered the wrong
password), causing confusion for the user.
Instead, Etcher will now display a nice "No polkit authentication agent
found." error dialog.
Change-Type: patch
Changelog-Entry: Display an error if no graphical polkit authentication agent was found.
See: https://github.com/jorangreef/sudo-prompt/pull/29
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Change-Type: patch
Changelog-Entry: Fix internal removable drives considered system drives in macOS Sierra.
See: https://github.com/resin-io-modules/drivelist/pull/86
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We were currently throwing a validation error if the error code was not
string when setting the flash results, however a number error code makes
sense in some cases, and its what its returned by a `ChildProcess`
object.
Change-Type: patch
See: https://github.com/resin-io/etcher/issues/609
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, the client application knows too much about how the flash
results are stored in the internal state, and relies on its structure to
perform its logic.
This PR introduces several getters to `FlashStateModel` and makes
`FlashStateModel.getFlashResults()` private, ensuring clients don't
depend on the flash results object.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
`lib/gui/app.js` contains a lot of code that should have been split long
ago. This PR extracts the "main page" logic into an actual page
component in `lib/gui/pages`.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
`ImageWriterService` currently has two responsibilities. It contains
logic to start and manage a flash process, and provides an API to
interact with the current flash state.
To honour the single responsibility principle, we extract
`FlashStateModel` from `ImageWriterService`.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We recently introduced a feature where a single available drive would be
auto-selected even if the user didn't select an image.
This introduces a potential scenario that brings the state to an
incoherent state:
- Start Etcher.
- Let a drive get auto-selected.
- Select an image larger than the auto-selected drive.
In this case, the drive is known to not have enough space to hold the
image, but remains auto-selected.
This PR makes sure the drive is deselected automatically in the above
scenario.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, we subscribe to the store change events and manually update a
`.state` property in `ImageWriterService`. A much more elegant approach
is to provide a `.getFlashState()` function that fetches the flash state
directly from the store when needed.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, TrackJS would receive errors when running Etcher locally.
Most of the errors reported from those environments are due to
developing Etcher, and create a lot of noise in TrackJS.
With this PR, we ensure Etcher will only report errors when running
inside an `asar` archive, which means its a "packaged" Etcher version.
Change-Type: patch
Changelog-Entry: Only enable error reporting if running inside an `asar`.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Currently, we intentionally prevent a drive from being auto-selected if
there was no selected image for UX purposes, however this way of
thinking has been challenged, and we didn't find a real UX problem by
doing this.
As a minor design improvement, we gray out the drive name in the main
window when an image hasn't been selected yet.
Change-Type: minor
Changelog-Entry: Perform drive auto-selection even when there is no selected image.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
When the user clicks on the backdrop of a modal, the modal promise gets
rejected, therefore triggering our application error handler.
UI Bootstrap provides a way to disable the backdrop completely, but
doesn't provide a way to allow a backdrop click to simply close the
modal rather than rejecting it, as if an issue happened.
To mitigate this issue, and still preserve the backdrop functionality,
we created `ModalService`, which abstracts the messy details of calling
`$uibModal`, and has custom logic to ignore "backdrop click" errors.
Change-Type: patch
Changelog-Entry: Fix "backdrop click" uncaught errors on modals.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
This version contains the following fixes:
- Correctly handle mount points with spaces in GNU/Linux.
- Correctly report drive sizes in macOS Sierra.
Change-Type: patch
Changelog-Entry: Upgrade `drivelist` to v3.2.4.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Instead of storing the whole selected drive object, we barely store a
reference to the corresponding drive in the available drives array (the
reference being the drive device).
This greatly simplifies the application state in the following ways:
- The drive metadata (size, description, etc) is not duplicated in the
state, enforcing a single source of truth.
- If the selected drive stops being available (e.g: is unplugged), the
reference doesn't hold anymore, making this functionality very natural
to implement.
- Makes `SelectionStateModel.isCurrentDrive()` much more inuitive, since
we don't have to document that changes in the metadata of the drive
object, or extra keys such as `$$hashKey` don't change the result of
this function.
- Ensures the state never goes into a problematic state where we try to
to write to an unavailable drive.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Consider a directory name such as `Juan's Files`. When passing such path
as an argument to the child writer proxy script, Bash will complain
with:
> Unexpected EOF while looking for matching `'`
The solution is to manually escape quotes on the image path.
Change-Type: patch
Changelog-Entry: Escape quotes from image paths to prevent Bash errors on GNU/Linux and OS X.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
The following PRs add support for a custom `_info` directory containing
metadata such an image URL, display name, logo, etc in
`etcher-image-stream`:
- https://github.com/resin-io-modules/etcher-image-stream/pull/16
- https://github.com/resin-io-modules/etcher-image-stream/pull/14
- https://github.com/resin-io-modules/etcher-image-stream/pull/13
Now that this module supports such metadata, we make use of it in the
GUI as follows:
- The file name is replaced with the display name.
- The file name links to the image URL.
- The "Select Image" logo is replaced with the image logo.
Some miscellaneous changes introduces in this PR to support the changes
described above:
- Implement `SelectionStateModel.getImageUrl()`.
- Implement `SelectionStateModel.getImageLogo()`.
- Implement `SelectionStateModel.getImageName()`.
- Ignore the "logo" image property when displaying the "Select image"
event, in order to not fill the console with SVG contents.
- Make `svg-icon` understand SVG strings as paths.
- Make `svg-icon` react to changes to the `path` attribute.
- Extract the core functionality of `openExternal` into
`OSOpenExternalService`.
- Upgrade `etcher-image-stream` to v3.0.1.
Change-Type: minor
Changelog-Entry: Support rich image extensions.
Fixes: https://github.com/resin-io/etcher/issues/470
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
The current error message is a bit blurry. The new one is much more user
friendly and directly guides the user towards a possible "next action".
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Versionist is a small utility to automatically generate CHANGELOGs based
on a certain commit convention.
This PR adds a new `changelog` npm script that calls Versionist to
preppend a new CHANGELOG entry based on our current structure.
Dealing with links to changes between versions in GitHub is hard to
automatise with a tool like Versionist. We chose to get rid of such
links since they are a nice to have, but don't provide any meaningful
value to maintainers, since they can manually see the diffs with `git
diff`.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
One of the steps incorrectly says "Re-install the dependencies. This
will update the `npm-shrinkwrap.json` file". The intention was to say
that you should `npm install` the *new* dependency, and that will update
the `npm-shrinkwrap.json` file.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
- Remove mention about Gitter channel covering Resin.io as a whole.
- Wrap text as the other documents.
- Move links to the bottom.
- Encourage users to submit OS/Etcher version and DevTools info.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We try our best to check that the images the user select are too big for
the selected drive as early as possible, but this probes to be
problematic with certain compressed formats, like bzip2, which doesn't
store any information about the uncompressed size, requiring a ~50s
intensive computation as a minimum to find it out.
For these kinds of formats, we don't perform an early check, but instead
gracefully handle the case where the drive doesn't have any more space.
This PR handles an `ENOSPC` error by displaying the alert orange ribbon,
and prompting the user to retry with a larger drive. This is a huge
improvement over the cryptic `EIO` error what was thrown before, and
over having Etcher freeze at a certain percentage point.
Change-Type: minor
Changelog-Entry: Display a nice alert ribbon if drive runs out of space.
See: https://github.com/resin-io/etcher/issues/571
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
The Etcher CLI doesn't care if the drive exists or not. The user will
eventually find out since the CLI will output an `ENOENT` when trying to
actually write data, however its nicer from a UX perspective to catch
this early on.
Change-Type: minor
Changelog-Entry: Validate the existence of the passed drive.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
This small entry explains how to upgrade a dependency and ensure
`npm-shrinkwrap.json` is updated correctly.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
Document that `parted` warnings about invalid GPT and physical block
size issues can be safely ignored as they are a consequence of "hybrid
iso's".
Fixes: https://github.com/resin-io/etcher/issues/312
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We were currently checking if a drive was large enough for an image by checking
the image file size, completely ignoring compression. This results on a small
chance of the uncompressed image not actually fitting in the drive, and
throwing more weird errors later on.
In order to mitigate this, we use the new `.getEstimatedFinalSize()` function
from `etcher-image-stream`, which returns the uncompressed size in the case of
compressed images.
We needed to update the build script for OS X to configure NPM correctly,
otherwise `lzma-native` throws some errors due to an ABI V8 incompatibility
issue.
Fixes: #571
See: https://github.com/addaleax/lzma-native/issues/25
Change-Type: patch
Changelog-Entry: Check if drive is large enough using the final uncompressed size of the image.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We were previously requiring users to run the exact same NodeJS version as
Electron bundled in order to avoid native add-on compilation incompatibilities,
however this problem is completely mitigated by configuring NPM to build
against an Electron specific V8.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
The dependencies are checked for in the build scripts, but for the user
convenience, we should also list them explicitly in the `CONTRIBUTING.md`
guide.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
This setting makes Etcher not filter non-removable drives, allowing you to
arbitrarily write to your system drives.
This is a dangerous option, therefore we present it in a separate section of
the settings page, and show an informative confirmation dialog.
Change-Type: minor
Changelog-Entry: Add an "unsafe" option to bypass drive protection.
Fixes: https://github.com/resin-io/etcher/issues/480
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
This version contains a fix to check the removable state of drives without a
file system in Windows.
Link: https://github.com/resin-io-modules/drivelist/blob/master/CHANGELOG.md
Change-Type: patch
Changelog-Entry: Upgrade `drivelist` to v3.2.2.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We were facing quite some issues that involved debugging the writer child
process, which proved to be a very difficult task given the lack of information
we're exposing to the parent.
This is an attempt to improve that situation.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
We're currently forcing a UNIX linebreak style, which causes hundreds of
complaints from the linter when cloning the project in a Windows machine
with a default configuration.
Signed-off-by: Juan Cruz Viotti <jviottidc@gmail.com>
