| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
The `lockfile_uses_separate_rubygems_sources` was causing a lockfile
incompatibility but in my opinion, this incompatibility is not necessary
in the general case.
|
| |
|
|
|
|
| |
Since we haven't even yet deprecated the old behavior.
|
|
|
|
| |
Bundler 2 can't be installed on these rubygems, so it's not necessary.
|
| |
|
| |
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
6971: Fix warning message about bundler version r=colby-swandale a=deivid-rodriguez
Fixes #6909.
### What was the end-user problem that led to this PR?
The problem was that we're recommending a command to make a warning go away that doesn't really make the warning go away.
### What was your diagnosis of the problem?
My diagnosis was that the current recommendation installs the latest bundler, which might not be the same as the version the lock file was created with. If it's not the same, the warning will not go away.
### What is your fix for the problem, implemented in this PR?
My fix is to recommend installing exactly the version the lockfile was created with.
### Why did you choose this fix out of the possible options?
I chose this fix because it's really simple and it should work.
Co-authored-by: David Rodríguez <deivid.rodriguez@riseup.net>
|
| |
| |
| |
| |
| |
| |
| | |
To actually make the warning go away, we need to recommend exactly
installing the locked version. Otherwise the latest version gets
installed, but not used to being a different major than the one locked,
so the warning stays.
|
| |
| |
| |
| | |
Since the current master will never run those rubies.
|
| | |
|
| |
| |
| |
| |
| | |
Deprecations enabled or not, we only want to print deprecations for the
running major version, never for future versions.
|
| |
| |
| |
| | |
Leave it as a setting with default true value.
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| | |
Including the version is confusing, in my opinion, because it's unclear
whether it refers to the future version of removal, or to the current
running version.
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
6974: Use newer list interface r=indirect a=deivid-rodriguez
### What was the end-user problem that led to this PR?
The problem was that our specs emit a lot of deprecations about the list command.
### What was your diagnosis of the problem?
My diagnosis was that we are using the deprecated interface to the list command everywhere.
### What is your fix for the problem, implemented in this PR?
My fix is to start using the new interface. If we don't use it ourselves, how could we be in a position to tell our users to start using it?
### Why did you choose this fix out of the possible options?
I chose this fix because we should do this, at same point, and doing it now simplifies enabling deprecations.
Co-authored-by: David Rodríguez <deivid.rodriguez@riseup.net>
|
| | | |
|
| | |
| | |
| | |
| | | |
So the we suggest an actually working command.
|
| |/ |
|
| |
| |
| |
| |
| | |
It break the examples of bundler. Because some examples detect the
different version of system ruby than test target version like trunk.
|
|/ |
|
|
|
| |
- See https://blog.travis-ci.com/2018-11-19-required-linux-infrastructure-migration for historical detail
|
|
|
|
| |
Before it was printing half a sentence to stderr and half to stdout.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
RbConfig::CONFIG["rubylibdir"].
https://bugs.ruby-lang.org/issues/15469
In ruby core, the bundled bundler was located same as rubylibdir.
When user used the specific version of default gems like json, Bundler
uses default gems under the rubylibdir, not specific version.
I avoid to add bundler directory from RUBYLIB enviromental variable,
Because its directory was already added $LOAD_PATH.
|
|\
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
6931: Add patch option in bundle config r=greysteil a=ankitkataria
### What was the end-user problem that led to this PR?
Issue #5994
### What was your diagnosis of the problem?
As mentioned by @indirect I added a `patch` option in the bundler config, which by default is false. If set to true, patch level update are preferred if no update levels are specified.
### What is your fix for the problem, implemented in this PR?
A patch level config option which can be opted into.
Co-authored-by: Ankit Kataria <ankitkataria28@gmail.com>
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
6315: Removed compatibility hack for old rubies. r=hsbt a=hsbt
### What was the end-user problem that led to this PR?
Nothing.
### What is your fix for the problem, implemented in this PR?
Bundler will never use them in version 2.
Co-authored-by: SHIBATA Hiroshi <hsbt@ruby-lang.org>
|
| |\ \ |
|
| | | | |
|
|\ \ \ \
| |_|_|/
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
6884: Remove unnecessary condition r=hsbt a=deivid-rodriguez
### What was the end-user problem that led to this PR?
The problem was that some code in the feature flags implementation seemed uncovered.
### What was your diagnosis of the problem?
My diagnosis was that the code should be either covered or removed.
### What is your fix for the problem, implemented in this PR?
My fix was to remove the code.
### Why did you choose this fix out of the possible options?
I chose this fix because the scenario where that code would get covered is more likely to be unintentional and risk introducing a bug. Defining a feature flag without a default block is the same as don't defining a feature flag and just relying on the default value for the setting. In these circumstances, it's more likely that the developer forgot to include a default block, so letting that case crash would be better, I think.
Co-authored-by: David Rodríguez <deivid.rodriguez@riseup.net>
|
| | |/
| |/|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This is currently dead code because all feature flags specify a default
block. With the current implementation, if we ever defined a feature
flag without specifying a default block, it would instead return the
default setting value. I think it is better to raise an error in that
case, since it would probably be an overlook and not something
intentional.
|
| | | |
|
|/ / |
|
|\ \
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
6856: Test against Ruby 2.6 and RubyGems 3 r=colby-swandale a=segiddins
### What was the end-user problem that led to this PR?
The problem was we weren't testing our compatibility with the latest and greatest.
Co-authored-by: Samuel Giddins <segiddins@segiddins.me>
Co-authored-by: Colby Swandale <me@colby.fyi>
|
| | | |
|
| | | |
|
|\ \ \
| |/ /
|/| |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
6843: Improve `clean_env` deprecation path r=deivid-rodriguez a=deivid-rodriguez
### What was the end-user problem that led to this PR?
The problem was that I noticed that there's still some use cases where the current implementation of `Bundler.with_clean_env` can be useful. Whereas the most common use case is to go back to the original environment before bundler was loaded, sometimes you actually want a fully "unbundled" environment. For example, I wanted to test a rails template in isolation (`rails new my_app -m my_template.rb`) in a library where we specifically set `BUNDLE_GEMFILE` in CI. In that context, `with_original_env` won't do the trick for me, because to properly test my template i need to make sure no bundler environment is set at all.
### What was your diagnosis of the problem?
My diagnosis was that we should probably undeprecate `(with_)clean_env`. But @indirect suggested that we should still deprecate the method name, because it's confusing what each developer understands by "clean". That's a very good point.
### What is your fix for the problem, implemented in this PR?
My fix is to move the functionality of `(with_)clean_env` to `(with_)unbundled_env`, and deprecate `(with_)clean_env`.
In addition, I noticed that the current deprecation message for `with_clean_env` actually mentioned the method `clean_env`, which is not a method the end user is using directly. So I changed the deprecation messages to suggest the proper alternative for each case (suggest `unbundled_env` as a replacement for `clean_env`; and `with_unbundled_env` as a replacement for `with_clean_env`).
### Why did you choose this fix out of the possible options?
I chose this fix because it undeprecates the functionality previously provided by `clean_env`, while still deprecating the `clean_env` name because of being confusing.
Co-authored-by: David Rodríguez <deivid.rodriguez@riseup.net>
|
| | |
| | |
| | |
| | |
| | | |
Previously the deprecation would point to the `Bundler.clean_env` method
even if not directly used by the end user.
|
| | |
| | |
| | |
| | | |
As a replacement to `with_clean_env`.
|
| | | |
|
| | |
| | |
| | |
| | | |
And thus, deprecate only the name, but not the functionality.
|
| | | |
|