dbDelta is unforgiving about field declarations

[SidHarrell]

the variable type is case sensitive:
int(22) != INT(22)

the mysql type BOOL or BOOLEAN comes back from the db as tinyint(1):
tinyint(1) != BOOLEAN

Not a huge issue, just annoying. Makes dbDelta fire off unnecessary sql.

[SidHarrell]

Also annoying is the fact that in our schema, we have to put 2 spaces in between PRIMARY KEY and the (:

PRIMARY KEY  (
VS
PRIMARY KEY (

This one is more problematic, since the core schema is using the 2 space rule, which would be easy enough to fix with a search and replace, but the patch I have for it would also break every plugin out there that uses the 2 space rule.

[SidHarrell]

I fixed that second patch to allow for single or double spaces between PRIMARY KEY and the ( so

PRIMARY KEY (

will work, but not break backwards compatibility with

PRIMARY KEY  (

[scribu]

+1 on the general idea.

Note that patch names should generally contain the ticket number, instead of some other arbitrary number.

Also, it's not clear which patches are meant to be final and which are just intermediary.

[SidHarrell]

Sorry about that. I'm new to the whole process.

upgrade_1589.patch (I was referring to the file and line#, sorry) is the patch for the first issue, to do a case insensitive comparison of fieldnames and convert the synonyms for boolean to tinyint(1).

upgrade_1651.patch breaks backwards compatibility, so ignore that one.

upgrade_1571.patch allows for single or double spaces after PRIMARY KEY

Thanks!

[nacin]

This was how dbDelta was originally designed. Setting as an enhancement.

[chriscct7]

@pento any thoughts?

[pento]

Before we even touch this, we need #29020 to land. :-)