enum is broken when multiple values and `_prefix/_suffix: true` options
are passed.
Because the `enum_prefix/enum_suffix` variables are overwritten from `true` to
`name` during the first loop, and one should return `true` at the second loop, got already overwritten.
Introduced by https://github.com/rails/rails/pull/40992
* Initial upload
* Moving toward a first pass
* Add Tables
I added tables for the configuration files/folders that are automatically generated.
I also added a table for the specific integrations. This way we can add basic links to the specific integration providing more detailed information about the integration.
* Add static assets section
* Bring over rails/webpacker development docs
* Respond to comments
* Update
* Update webpacker_guide.md
I think `include` should be `import` and would like to emphasis `import` over `require` in an effort to promote ES module syntax.
* Expand on image docs
I thought some additional details are needed to clarify image usage, especially from a Rails view.
* Update webpacker_guide.md
Add details for Webpacker + Docker section
* Update webpacker_guide.md
Make important distinction about pack files. Add example directory layout.
* Reword sentence
* Add section for Rails engines
* Rename webpacker guide
* Remove Docker section
* Add link to additional documentation
* Rework section formatting according to Rails guidelines
* Add link to webpack documentation
* Add note about NPM and Yarn
* Update guides/source/webpacker.md
Fix typo
Co-authored-by: Karl Entwistle <me@karlentwistle.com>
Co-authored-by: Noel Rappin <noel.rappin@joinroot.com>
Co-authored-by: Niklas Häusele <niklas.haeusele@gmail.com>
Co-authored-by: Tasos Latsas <tlatsas2000@gmail.com>
Co-authored-by: Noel Rappin <noelrappin@gmail.com>
Co-authored-by: Karl Entwistle <me@karlentwistle.com>
f72f743 introduces truncate(scale) in the Numericality validator.
This behaviour conflicts with AR decimal type conversion,
which uses round(scale) instead.
Changes the Numericality validator in order to use
round(scale) for consistency.
If you're using `strict_loading` in an application the previous message
only told you what class was lazily loaded. This change updates the
error message to include both the class and the association name. This
is useful because now you won't need to lookup the association name in
the application, the error message will tell you exactly which symbol
preload is missing.
In our recent optimizations to route matching performance we introduced
an issue with routes that had an optional segment which wasn't separated
by a "." or "/".
Similar to #40809 which is fixed by #40815, but it is caused by dbda5fe.
`value.id if value.respond_to?(:id)` is not always correct if primary
key is customized on the belongs_to association.
Use `reflection.join_primary_key` in that case.
Fixes#41081.
The ability has lost due to reverted #39321 in #41049.
We should allow updating with dirty locking value to work the documented
usage, but if casted value has no difference (i.e. regarded as no dirty),
identify the object by the original (uninitialized default) value.
When using `Pathname#relative_path_from`, the receiver and the argument
must either both be absolute or both be relative:
https://ruby-doc.org/stdlib-2.7.2/libdoc/pathname/rdoc/Pathname.html#method-i-relative_path_from
> If self is absolute, then base_directory must be absolute too.
>
> If self is relative, then base_directory must be relative too.
If `file_fixture_path` is a relative path and `fixture_path` is an
absolute path, this line would previously raise an ArgumentError.
