composer.json - Relax psr/log constraint. Improve D8 compatibility.

[totten]

Overview

This addresses a composer conflict that's reported when trying to install
on Drupal 8.7. For example, run these commands:

drush8 dl drupal-8.7.x
cd drupal-8*
composer require civicrm/civicrm-core:5.22.x-dev

This is a port of #16470 for master.

Before

The install fails because civicrm-core requires psr/log:~1.1, and something else
is prodding us to use psr/log:1.0.2.

./composer.json has been updated
Loading composer repositories with package information
Updating dependencies (including require-dev)
Your requirements could not be resolved to an installable set of packages.

  Problem 1
    - Installation request for civicrm/civicrm-core 5.22.x-dev -> satisfiable by civicrm/civicrm-core[5.22.x-dev].
    - Conclusion: remove psr/log 1.0.2
    - Conclusion: don't install psr/log 1.0.2
    - civicrm/civicrm-core 5.22.x-dev requires psr/log ~1.1 -> satisfiable by psr/log[1.1.0, 1.1.1, 1.1.2, 1.1.x-dev].
    - Can only install one of: psr/log[1.1.0, 1.0.2].
    - Can only install one of: psr/log[1.1.1, 1.0.2].
    - Can only install one of: psr/log[1.1.2, 1.0.2].
    - Can only install one of: psr/log[1.1.x-dev, 1.0.2].
    - Installation request for psr/log (locked at 1.0.2) -> satisfiable by psr/log[1.0.2].

Installation failed, reverting ./composer.json to its original content.

After

It should work. However, this is hard to demonstrate via r-run without merging.

Comments

The substantive differences between psr/log in v1.0 and v1.1 relate to LoggerInterfaceTest and TestLogger:

php-fig/log@1.0.2...1.1.2

However, civicrm-core does not use LoggerInterfaceTest or TestLogger, so it shouldn't matter.

For the standard tarballs which use composer.lock, this does have the side-effect of bumping up from 1.1.0 to 1.1.2.

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[eileenmcnaughton]

Seems fine for master - not sure why it's against 5.22 though?

[totten]

Seems fine for master - not sure why it's against 5.22 though?

Yeah, this is not a regression AFAIK, but it is an installability issue. So I wouldn't put it against the stable - but the RC is a maybe.

I'll be OK if it's just master.

Merging into 5.22 as well would mean that there's one less crufty bit in some of the CI scripts.

[eileenmcnaughton]

Well master will be the rc tomorrow .... & then it will get a full cycle of testing etc

[seamuslee001]

Yeh i'm good for master i don't think its crucial for 5.22 tbh merging this and reviewed the changes and i don't think we will be harmed by relaxing the requirements here