Prevent pgsql driver from trying to implode primary key fields that are defined by an array [#2181283]

This issue has been split from the meta/parent issue #2001350: [meta] Drupal cannot be installed on PostgreSQL.
Updated: Comment #2

Problem/Motivation

Prevent pgsql driver from trying to implode primary key fields that are defined by an array (column, key_length) as that syntax is not supported. See Comment #75.

This change to primary key schema syntax was made as part of #731724: Convert comment settings into a field to make them work with CMI and non-node entities .

Proposed resolution

Implement createPrimaryKeySql() in core/lib/Drupal/Core/Database/Driver/pgsql/Schema.php to account for new primary key schema syntax.

Remaining tasks

Create a smaller patch from the patch in the parent issue.
This patch cannot be tested in PostgreSQL until the transaction bug is fixed because Drupal cannot be installed. However the bug has a known and working fix from the parent issue.
Create follow-up issue to fix method names in pgsql driver to match standard per Crell's comment.

API changes

New method createPrimaryKeySql() which matches other database drivers.

Comment	File	Size	Author
#2	drupal-2181283-pgsql-primary-key-1.patch	2.19 KB	mradcliffe

Comments

Comment #1

mradcliffe

he/him

English

commented 25 January 2014 at 02:59

Priority:

Major

» Critical

Comment #2

mradcliffe

he/him

English

commented 25 January 2014 at 19:12

Issue summary:	View changes
Status:	Active	» Needs review
Related issues:		+#731724: Convert comment settings into a field to make them work with CMI and non-node entities

Status	File	Size
new	drupal-2181283-pgsql-primary-key-1.patch	2.19 KB

Created patch with resolution from parent issue, and removed the leading underscore from the method name.

Added more information to issue summary about the cause of the issue.

Comment #3

bzrudi71 commented 4 February 2014 at 11:19

Just applied patch and it works as expected. As this patch only affects the PostgreSQL driver itself I see no side issues with that, so we should get this in. Especially as PostgreSQL testbot is now up and running and the issue from PostgreSQL failed to connect seems gone.