The error recovery process is necessary, but is particularly subtle. An
error may occur basically any time some user code is called, which may
leave the internal owl state in an invalid situation.
In this commit, we try to improve Owl behaviour when a crash occurs in a
onMounted hook. Before this commit, it was possible for other component
mounted hooks to be called even though the component was not mounted, or
for willUnmount hooks to be called even though the component was
mounted, but its onMounted hooks had not been called.
All these situations need to be handled when we catch an error. In this
commit, we try to clean more internal state in the error handling code,
to make Owl more consistent.
If a component onMounted hook causes an error, and an error handler
tries to reinitiate a new correct render, the fiber may be recycled. But
the scheduler removes it from its own set of tasks, so it will be
ignored.
We simply check for this condition before removing it from the tasks
set.
Previously, when a template failed to compile because of a syntax error
(typically because we don't do any syntax checking when compiling
expression, allowing invalid expressions to be transpiled successfully),
the error reporting was very minimal: you would only get the error
message itself (eg: "Unexpected token") with the stack information of
the error pointing to the call to `new Function` in owl, which is not
very useful.
This commit catches the compilation error and completes it with
information about the template name when available, and also adds the
generated code to the error message, allowing the user to just
copy/paste it in their web console or code editor to get a more precise
location for the error.
Before this commit, most of the time, components are destroyed when the
virtual dom is patched and a component node is removed. However, since
Owl is asynchronous and a component may take some time to get ready
(with onWillStart), it can happen that a component is created, then
before it is ready, it is recreated. In that case, the initial instance
has to be destroyed.
Before this commit, the destroy operation was done immediately, when we
cancel the current fibers. However, this means that we cannot have a
guarantee between micro task ticks that a component has not been
destroyed in the meantime.
For example, in Odoo, it is common to use the rpc service, which will
throw an error if called by a destroyed component. But because of the
possible destruction of a component at any time, the following code is
unsafe:
async loadSomeData() {
// guaranteed to be called when component is alive
await Promise.resolve();
// however here, component may have been destroyed
this.rpc(...)
}
So, to prevent this issue, we can slightly delay the destroy operation.
It is not entirely trivial, since we need to find a way to neutralize
the component in the meantime. But it seems like performing all that
kind of operation at the "commit" phase (so, the request animation frame
callback) makes sense to me.
So, this commit modifies the code to add a new component status
(cancelled) and use it to cancel components that are waiting to be
destroyed. These components will then be destroyed as soon as the
requestanimation frame starts, before all other dom operations.
This commit makes it so that when an error occurs in an owl app and none
of the registered error handlers are able to handle it, we rethrow the
error instead of just logging it to the console and swallowing it. This
allows users of owl to handle errors that happen in owl applications by
using event listeners for error and unhandledrejection events on the
window.
Previously, when wrapping errors in wrapError, if the error was not an
actual error object, we wouldn't set the cause property on the wrapping
error correctly. The "instanceof Error" check is simply there so that we
can know whether we can add the original errors message to the wrapping
error, but the line that sets the error's cause was mistakenly moved
into that condition.
This commit also fixes the wrapping error's message in the case of
non-Error objects, to avoid having "the following error occurred in
hookname:" with nothing after the colon which is confusing/misleading.
This commit makes all errors thrown in owl use a custom error class. The
main point of this is to always wrap user-code errors that happen during
the owl lifecycle so that they can be treated uniformly in onError by
checking the cause property, and also allows user code to differenciate
owl errors from non-owl errors reliably at runtime.
Before this commit, wrapping an error occurring in async code
would result in an unhandledpromise exception, because we created
another promise, that would be rejected and that we didn't catch.
If you declare a child component which is not actually a Component,
the error message is not very friendly and not very helpfull to find
what happens and which child component is wrong.
```
const ChildComponent = "not a component constructor";
class MyComponent extends Component {
static components = { ChildComponent };
}
```
This commit improves the type declaration for those working with Typescript
and adds a runtime check for javascript codebases
Previously, we were wrapping the entire renderFn in a try/catch, causing
errors during template execution to be caught and wrapped by
onWillRender/onRendered which is undesirable. Now we only wrap the hook
that's being registered.
Previously, a crash in a lifecycle hook for any reason would throw an
error whose stack trace started from the scheduler and contained only
the place where the hook was called by owl, but not the place where the
hook was registered by the user. This proved very difficult for users to
debug as they cannot really tell which component registered that hook.
This commit alleviates the issue by creating a new Error when the hook
is originally called, and wrapping the registered callback in a try
catch, throwing an error with the correct stack trace instead of the
error in the hook, and setting the error in the hook as the cause of
this synthetic error.
Owl provides a way to manage errors occuring in component lifecycle
methods. However, before this commit, these errors were not always
logged or visible, which is very annoying in the common developer
workflow (doing something, checking it works, seeing no error but a
broken interface).
In this commit, we make sure errors are logged/throws in all cases:
- if an error occurs in a mounting operation => the promise is rejected
(which will log the error)
- if an error occurs after the mounting operation and is not handled by
any error handlers => the error will be logged (with console.error).
Also, in that case, this commit adds a warning to explain that owl
destroys the root component, which will help developers understanding
what happened.
Before this commit, class inheritance when using the onError hook was unclear nay wrong.
After this commit, error handlers are called from the bottom up in the inheritance hierarchy.
If a handler doesn't rethrow the error, the handling stops there and no other handler is called.
If a handler does rethrow, the handlers declared in a parent class are executed.
In the following situation: A parent of B, B parent of C, with an error
when C is mounted, caught by B and retriggering a rendering in B, then
the onMounted hook of A wasn't properly called. This commit fixes this
problem.
Before this commit, errors triggered at the level of the fiber (as opposed to at the level
of a component's rendering), were handled as the very top level of the rendering, that is,
in the scheduler.
This was wrong because components below in the rendering tree would not have a chance to handle their
children's or their own errors.
After this commit, error triggered in willPatch, onMounted and onPatched are correctly handled
at the closest component to where they were thrown.
Have a Child Compnent which has one component that succeeds and another
one that fails at its instanciation.
The Child component handles the Errors by rendering itself.
Before this commit, the error handling algorithm made impossible for the scheduler to finish.
This was because the current fiber was still counted as ongoing, when it was actually completed.
After this commit, this use case is handled correctly.
This commit reintroduces some tests for the t-set directive and make
them pass. For that, it was necessary to adapt the qweb compiler in
order to get the following behaviors:
A t-set can affect parent contexts (up to the first parent tagged as
boundary) when the key changed is found in one of the parent contexts.
Some context are marked as boundaries in such a way that
- rendering contexts (e.g. components) cannot be modified via a t-set.
- a t-set in a t-call body or in a called template can never change a
context above the t-call context.
Code prettification has been done.
Snapshots have been modified.
The point is to have visibility on the development of the owl2 features.
This commit reintroduces some tests keeping them skipped in order to fulfill that purpose.
There still are some missing tests though.
Nicolas Bayet