Adds duplicate sequences to schema.rb

[mxey]

When using pg_sequencer, the primary key sequences are also added to schema.rb, but they are automatically created as part of create_table. Running rake:db:setup fails because Rails tries to create sequences foo_id_seq that already exist.

[tonyc]

Hi Max,

Any thought on what the proper behavior should be?

Our usecase was that we were always manually creating the sequences and PK fields for tables, not relying on a column actually listed as PRIMARY KEY (don't ask :)).

I'm wondering if there's any way we can tell whether a sequence is a PK sequence, and if so, skip it so we don't conflict with create_table. I haven't looked at the code in a while, but I think we just query for all sequences.

On Sat, Sep 15, 2012 at 6:58 PM, Maximilian Gaß notifications@github.comwrote:

When using pg_sequencer, the primary key sequences are also added to schema.rb, but they are automatically created as part of create_table. Running rake:db:setup fails because Rails tries to create sequences foo_id_seq that already exist.

— Reply to this email directly or view it on GitHubhttps://github.com/code42/pg_sequencer/issues/1.

[mxey]

Can you give me your schema.rb as an example? I would like to compare it to mine to see how your tables look in that definition.

[tonyc]

Now that I think of it, we never actually used the schema dumper feature. We ended up using SQL schema dumps, so the ruby version of schema.rb is essentially untested, like a boss :)

Would it be easy for you to pastebin a sample schema.rb? I'm not actively involved with the project that used pg_sequencer anymore, but maybe this is a good excuse to get back involved.

On Sat, Sep 29, 2012 at 4:50 PM, Maximilian Gaß notifications@github.comwrote:

Can you give me your schema.rb as an example? I would like to compare it to mine to see how your tables look in that definition.

— Reply to this email directly or view it on GitHubhttps://github.com/code42/pg_sequencer/issues/1#issuecomment-9008635.

[mxey]

I gisted both the dumped schema.rb and the one where I stripped out duplicate sequences: https://gist.github.com/3829365

[tonyc]

OK. So it does boil down to not simply writing out all of the sequences to schema.rb. Assume that sequences will be generated for calls to create_table, which means filtering out sequences that belong to the PK of a table.

Does that sound right?

On Wed, Oct 3, 2012 at 2:46 PM, Maximilian Gaß notifications@github.comwrote:

I gisted both the dumped schema.rb and the one where I stripped out duplicate sequences: https://gist.github.com/3829365

— Reply to this email directly or view it on GitHubhttps://github.com/code42/pg_sequencer/issues/1#issuecomment-9119524.

[mxey]

On 03.10.2012, at 22:54, Tony Collen notifications@github.com wrote:

OK. So it does boil down to not simply writing out all of the sequences to schema.rb. Assume that sequences will be generated for calls to create_table, which means filtering out sequences that belong to the PK of a table.

Does that sound right?

Yes.

[achempion]

I wrote a quick fix to solve problem. Read rails section for explanation - https://github.com/achempion/pg_sequencer

[benlinton]

Here's a robust solution: https://github.com/code42/pg_sequencer/pull/11

This approach dumps all sequences to the schema.rb unless they're owned by any table's primary key.

[amoravec]

Thank you for your feedback, unfortunately this project was effectively abandoned and will no longer be maintained by Code42.