This PR was merged into the 4.3 branch.
Discussion
----------
[Messenger] Show exceptions after multiple retries
| Q | A
| ------------- | ---
| Branch? | 4.3
| Bug fix? | yes
| New feature? | no
| BC breaks? | no
| Deprecations? | no
| Tests pass? | yes
| Fixed tickets | #32311
| License | MIT
| Doc PR | n/a
After retrying a failed message, the `RedeliveryStamp` looses it's exception information. This PR will remedy that.
Commits
-------
598bd92313 [Messenger] Show exceptions on after empty fails
This PR was merged into the 4.3 branch.
Discussion
----------
Revert "[Messenger] Fix exception message of failed message is dropped
| Q | A
| ------------- | ---
| Branch? | 4.3
| Bug fix? | yes
| New feature? | no <!-- please update src/**/CHANGELOG.md files -->
| Deprecations? | no <!-- please update UPGRADE-*.md and src/**/CHANGELOG.md files -->
| Tickets |
| License | MIT
| Doc PR |
This reverts #33600 because it makes the message grow for each retry until AMQP cannot handle it anymore. On each retry, the full exception trace is added to the message. So in our case on the 5th retry, the message is too big for the AMQP library to encode it. AMQP extension then throws the exception
> Library error: table too large for buffer
(ref. https://github.com/alanxz/rabbitmq-c/issues/224 and https://github.com/pdezwart/php-amqp/issues/131) when trying to publish the message.
To solve this, I suggest to revert #33600 (this PR) and merge #32341 instead which does not re-add the exception on each failure.
Btw, the above problem causes other problematic side-effects of Symfony messenger. As the new retry message fails to be published with an exception, the old (currently processed message) also does not get removed (acknowledged) from the delay queue. So rabbitmq redelivers the message and the same thing happens forever. This can block the consumers and have a huge toll on your service. That's just another case for https://github.com/symfony/symfony/issues/32055#issuecomment-529630654. I'll try to fix this in another PR.
Commits
-------
3dbe924f47 Revert "[Messenger] Fix exception message of failed message is dropped on retry"
This PR was merged into the 4.4 branch.
Discussion
----------
[Validator] Set Length::$allowEmptyString to false when a NotBlank contraint is defined
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Since #31528, we are told to do this kind of changes to our entities:
```diff
/**
* @Assert\NotBlank()
- * @Assert\Length(min="10")
+ * @Assert\Length(min="10", allowEmptyString=false)
*/
public $description;
```
But the `NotBlank` already says it - this is just boilerplate. More critically, this also means we cannot write annotations that are compatible with 3.4, 4.4 and 5.0 at the same time, making FC/BC hard.
By setting `Length::$allowEmptyString` to `false` when a `NotBlank` contraint is defined, we fix both issues.
/cc @ogizanagi
Commits
-------
840f7e7d88 [Validator] Set Length::$allowEmptyString to false when a NotBlank contraint is defined
This PR was merged into the 4.4 branch.
Discussion
----------
[FrameworkBundle] Don't reset the test container but the real one instead
| Q | A
| ------------- | ---
| Branch? | 4.4 for features / 3.4 or 4.3 for bug fixes <!-- see below -->
| Bug fix? | yes/no
| New feature? | yes/no <!-- please update src/**/CHANGELOG.md files -->
| Deprecations? | yes/no <!-- please update UPGRADE-*.md and src/**/CHANGELOG.md files -->
| Tickets | Fix #... <!-- prefix each issue number with "Fix #", if any -->
| License | MIT
| Doc PR | -
After #31202 and #32056, the tearDown method keeps throwing deprecation notices about "Getting the container from a non-booted kernel". The reason is that resetting the test-container calls `$kernel->getContainer()` while the kernel has been shut down.
This fixes it and a few other glitches found meanwhile.
Commits
-------
8e16143256 [FrameworkBundle] Dont reset the test container but the real one instead
This PR was merged into the 4.4 branch.
Discussion
----------
[SecurityBundle] test with doctrine-bundle 2
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Commits
-------
e3261f4f7f [SecurityBundle] test with doctrine-bundle 2
This PR was merged into the 4.4 branch.
Discussion
----------
[Debug] remove return types that break FC badly
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
The return type on Debug's FlattenException blocks creating a child class that is compatible with both v4.4 and v5.0.
Removing it fixes the issue with no BC break.
Adding `final` on `setPrevious` will allow updating its type hint in v5.0.
Commits
-------
cb5ef6ec18 [Debug] remove return types that break FC badly
This PR was merged into the 4.4 branch.
Discussion
----------
Add .gitignore to .gitattributes
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | #33946
| License | MIT
Commits
-------
246c5fdf43 Add .gitignore to .gitattributes
This PR was merged into the 4.4 branch.
Discussion
----------
[Mailer] Fix typo
| Q | A
| ------------- | ---
| Branch? | 4.4 for features / 3.4 or 4.3 for bug fixes <!-- see below -->
| Bug fix? | yes
| New feature? | no <!-- please update src/**/CHANGELOG.md files -->
| Deprecations? | no <!-- please update UPGRADE-*.md and src/**/CHANGELOG.md files -->
| Tickets | Fix#34006
| License | MIT
| Doc PR | n/a
The XML is different depending on the way we are sending email. So, it's `SendEmailResult` when using the API and `SendRawEmailResult` when using the HTTP class (we are then sending the raw email).
Commits
-------
4bd7cb0368 [Mailer] Fix SES Message Id retrieval
This PR was merged into the 4.4 branch.
Discussion
----------
[ExpressionLanguage][Lexer] Exponential format for number
Exponential format has been added for numbers.
Ex: 1.99E+3 === 1990,
Ex: expression (1 + 1.99E+3) = 1991
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | yes
| Deprecations? | no
| Tickets |
| License | MIT
| Doc PR |
Exponential format has been added for numbers.
Ex: 1.99E+3 === 1990,
Expressions:
0.1e+2 = 10
1e-2 = 0.01
(1 + 1.99E+3) = 1991
and etc...
Commits
-------
430ec32992 [ExpressionLanguage][Lexer] Exponential format for number
This PR was merged into the 4.4 branch.
Discussion
----------
Use port 465 for SES SMTP transport
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | Fix#34064
| License | MIT
Commits
-------
8492e260cb Use port 465 for SES SMTP transport
This PR was merged into the 4.4 branch.
Discussion
----------
[Messenger] Fix worker-only Doctrine middleware from running always
| Q | A
| ------------- | ---
| Branch? | 4.4 (or 4.3?, this is a bug fix)
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | Fix#32436 Depends on #34069
| License | MIT
| Doc PR | not needed
Several Doctrine middleware are only meant to be run in a "worker" context: we want to "ping" the connection, "close" the connection and "clear" the entity manager ONLY when we are receiving messages. Before this PR, it was done always, which causes bad behavior for sync messages (imagine your Doctrine connection being closed in the middle of a controller or see https://github.com/symfony/symfony/pull/31334#issuecomment-544288437).
This fixes that in a pragmatic way: no new system for "worker-only" middleware or anything like that: just make the middleware smart enough to only do their work when a message is being received async.
Commits
-------
290a72917b Fixing issue where worker-only middleware were run in all contexts
This PR was merged into the 4.4 branch.
Discussion
----------
[ErrorRenderer] FlattenException cannot be final
| Q | A
| ------------- | ---
| Branch? | master
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Making it final forces tight coupling with the implementation as soon as one type-hints for the class.
That's a blocker on making e.g. `EasyAdminBundle` compatible with Symfony 5.
Commits
-------
b125835056 [ErrorRenderer] FlattenException cannot be final
This PR was merged into the 4.4 branch.
Discussion
----------
[Messenger] Removing "sync" transport and replacing it with config trick
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no (but removes some hackiness)
| New feature? | no
| Deprecations? | yes (BC break yes)
| Tickets | Helps #34066
| License | MIT
| Doc PR | not needed
I introduced `SyncTransport` in 4.3 as a way to allow people to, for example, route a message to an `async` transport and then override that to a `sync` transport in the dev environment. The implementation was always a bit of a hack, and makes this one transport behave different than others *and* makes messages routed to this transport behave a *bit* different than sync messages that are simply not routed at all.
Thanks for an idea from Nicolas, this fixes that. Gone are `SyncTransport`. Instead, the `sync://` transport string is still supported, but it's just a "config" trick: messages routed to transports that are "sync" are simply filtered out of the "senders" in the DI extension. Nice & clean.
Commits
-------
3d4e59a10b [Messenger] Removing "sync" transport and replacing it with much nicer config trick
This PR was merged into the 4.3 branch.
Discussion
----------
[HttpClient] skip tests implemented in 4.4
| Q | A
| ------------- | ---
| Branch? | 4.3
| Bug fix? | no
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Needed after #34051 and #34044
Commits
-------
ae86ab18fa [HttpClient] skip tests implemented in 4.4
This PR was merged into the 4.4 branch.
Discussion
----------
[VarDumper] improve displaying cut closures
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | no
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Minor but still an improvement so 4.4: this change makes closures replaced by `CutStub` be displayed with their signature instead of just the `Closure` label.
Commits
-------
2b0a11de02 [VarDumper] improve displaying cut closures
This PR was merged into the 4.4 branch.
Discussion
----------
[Lock] Set ReturnType of LockFactory to LockInterface
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | Fix #...
| License | MIT
| Doc PR |
LockFactory I think should return a LockInterface and not the Lock class.
/cc @chalasr
Commits
-------
ff1fa57ef2 Set ReturnType of LockFactory to LockInterface
This PR was merged into the 3.4 branch.
Discussion
----------
[TwigBridge] do not render errors for checkboxes twice
| Q | A
| ------------- | ---
| Branch? | 3.4
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | Fix#34008
| License | MIT
| Doc PR |
Commits
-------
9eddea97d8 do not render errors for checkboxes twice
This PR was merged into the 4.4 branch.
Discussion
----------
[DI] made the `%env(base64:...)%` processor able to decode base64url
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | yes
| Deprecations? | no
| Tickets | -
| License | MIT
| Doc PR | -
Commits
-------
3d12e46692 [DI] made the `%env(base64:...)%` processor able to decode base64url
This PR was merged into the 4.3 branch.
Discussion
----------
[Messenger] Fix ignored options in redis transport
| Q | A
| ------------- | ---
| Branch? | 4.3
| Bug fix? | yes
| New feature? | no
| Deprecations? | no
| Tickets | Fix#33425
| License | MIT
| Doc PR | -
Also fixes redis authentication failure handling (inline with invalid db index handling, borrowed from symfony/cache).
/cc @alexander-schranz
Commits
-------
c83ff94c37 [Messenger] Fix ignored options in redis transport
This PR was squashed before being merged into the 4.4 branch.
Discussion
----------
[HttpClient] Add a canceled state to the ResponseInterface
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | yes
| Deprecations? | no
| Tickets |
| License | MIT
| Doc PR | none yet
As discussed in https://github.com/terminal42/escargot/pull/7 with @nicolas-grekas.
However, I'm not quite sure about the tests here because it looks like the `MockResponse` always throws a `TransportException` when I cancel a response, even if I call `$response->getContent(false)`. Is that a desired behaviour? :) Need some guidance here to finish that PR maybe.
Note: I've also sorted the info keys alphabetically and added their types because it feld incomplete and a bit random otherwise. Hope that's okay :)
Commits
-------
bbb8ed11ff [HttpClient] Add a canceled state to the ResponseInterface
This PR was merged into the 4.4 branch.
Discussion
----------
[FrameworkBundle] Add `secrets:*` commands and `%env(secret:...)%` processor to deal with secrets seamlessly
| Q | A
| ------------- | ---
| Branch? | 4.4
| Bug fix? | no
| New feature? | yes
| Deprecations? | no
| Tickets | Fix#27351
| License | MIT
| Doc PR | symfony/symfony-docs/pull/11396
This PR continues #31101, please see there for previous discussions. The attached patch has been fine-tuned on https://github.com/nicolas-grekas/symfony/pull/33 with @jderusse.
This PR is more opinionated and thus a lot simpler than #31101: only Sodium is supported to encrypt/decrypt (polyfill possible), and only local filesystem is available as a storage, with little to no extension point. That's on purpose: the goal here is to provide an experience, not software building blocks. In 5.1, this might be extended and might lead to a new component, but we'd first need reports from real-world needs. Having this straight-to-the-point in 4.4 will allow gathering these needs (if they exist) and will immediately provide a nice workflow for the need we do want to solve now: forwarding secrets from dev to prod using git in a secure way.
The workflow this will allow is the following:
- public/private key pairs are generated in the `config/secrets/%kernel.environment%/` folder using `bin/console secrets:generate-keys`
- for the prod env, the corresponding private key should be deployed to the server using whatever means the hosting provider allows - this key MUST NOT be committed
- the public key is used to encrypt secrets and thus *may* be committed in the git repository to allow anyone *that can commit* to add secrets - this is done using `bin/console secrets:set`
DI configuration can reference secrets using `%env(secret:...)%` in e.g `services.yaml`.
There is also `bin/console secrets:remove` and `bin/console debug:secrets` to complete the toolbox.
In terms of design, vs #31101, this groups the dual "encoder" + "storage" concepts in a single "vault" one. That's part of what makes this PR simpler.
That's all folks :)
Commits
-------
c4653e1f65 Restrict secrets management to sodium+filesystem
02b5d740e5 Add secrets management
8c8f62390a Proof of concept for encrypted secrets
