285443 - Schema::Pg should remove FULLTEXT attribute from indexes

Status: RESOLVED FIXED

(Bugzilla :: Bugzilla-General, defect, P2)

Description

[Max Kanat-Alexander]

In order to be consistent with what's actually on the disk in the database, the
Pg _adjust_schema, at this point, should probably remove the FULLTEXT option
from all of the FULLTEXT indexes. (This is important so that later we can detect
from the serialized schema that the indexes aren't FULLTEXT and upgrade them.)

Comment 1

[Max Kanat-Alexander]

Even better would be to do it in Schema::Pg::_initialize, so that we do it to
the abstract schema, too.

Comment 2

[Ed Sabol]

Attachment: V1

It should be done in Mysql::_initialize() after the call to
SUPER::_initialize().

I don't think you should be changing the abstract schema personally, but feel
free to take this (untested!) patch and run with it as you see fit, Max.

Comment 3

[Max Kanat-Alexander]

Comment on attachment 176892 [details] [diff] [review]
V1

Thank you, Ed. :-)

Yeah, I know that we're starting to do weird things with the Schema, but I had
a feeling that would happen once we actually had it. :-) It's really quite a
useful tool. :-)

Comment 4

[Max Kanat-Alexander]

By the way, thank you for that ULTRA-fast patch, Ed. :-) That was awesome.

Comment 5

[Ed Sabol]

What about removing FULLTEXT from the abstract schema? I presume that's for serialization purposes, 
so I guess you could implement that in bug 285113 if deemed necessary.

Comment 6

[Max Kanat-Alexander]

Comment on attachment 176892 [details] [diff] [review]
V1

Oh, we don't change the {abstract_schema}, though.

Comment 7

[Ed Sabol]

Attachment: V2

(In reply to comment #6)
> Oh, we don't change the {abstract_schema}, though.

Yeah, I was hoping to discuss that. It's easily enough rectified, but
I'm still coming to grips with the need to make the abstract schema
database-specific. Kind of goes contrary philosophically to the
original purpose. But I suppose it's necessary for your serialization
stuff, so here's another patch...

Comment 8

[Ed Sabol]

Attachment: V3

Oh, that one has tabs. (I can't seem to turn that off in the Emacs perl-mode
that I'm using.)

Comment 9

[Max Kanat-Alexander]

Comment on attachment 176903 [details] [diff] [review]
V2

Yeah, it's because of the serialization stuff.

Thanks, this looks good. :-)

Comment 10

[Max Kanat-Alexander]

Comment on attachment 176905 [details] [diff] [review]
V3

Oh, thanks for catching that. :-) Otherwise I would have had to fix it right
after checkin when the Tinderbox complained. :-)

Comment 11

[Max Kanat-Alexander]

Checking in Bugzilla/DB/Schema/Pg.pm;
/cvsroot/mozilla/webtools/bugzilla/Bugzilla/DB/Schema/Pg.pm,v  <--  Pg.pm
new revision: 1.3; previous revision: 1.2
done