27 lines
1.3 KiB
YAML
27 lines
1.3 KiB
YAML
|
decisions_log:
|
||
|
|
- question: >
|
||
|
|
Which image processing library will back JIT Generation — Glide or Intervention
|
||
|
|
Image?
|
||
|
|
decision: Glide
|
||
|
|
rationale: >
|
||
|
|
Chosen over Intervention Image as the JIT-generation backend (see
|
||
|
|
`media_manager_architecture.decisions_log` for the same decision applied there).
|
||
|
|
- question: >
|
||
|
|
Should Composer dependency versions be pinned to exact versions, or use caret/
|
||
|
|
tilde ranges, given the "no compatibility baggage" stance already taken on the
|
||
|
|
minimum PHP version?
|
||
|
|
decision: >
|
||
|
|
Use caret/tilde ranges, but make sure the minimums are set appropriately.
|
||
|
|
rationale: >
|
||
|
|
Caret/tilde ranges allow non-breaking dependency updates while still letting the
|
||
|
|
minimum-version floor for each package be set deliberately.
|
||
|
|
- question: >
|
||
|
|
Add a static analysis tool and a code-style tool to the stack to support the
|
||
|
|
"Quality First" pillar, since none was listed.
|
||
|
|
decision: >
|
||
|
|
Added `backend_static_analysis` (PHPStan for general code quality, Psalm for
|
||
|
|
code security — both are used, for different purposes) and
|
||
|
|
`backend_code_style_tool` (PHP CS Fixer) above.
|
||
|
|
rationale: >
|
||
|
|
PHPStan and Psalm catch different classes of issues (general type/quality bugs
|
||
|
|
vs. security-focused taint analysis), so both are kept rather than picking one.
|