On Sat, Dec 14, 2013 at 12:22 PM, Tom Lockhart <tlockhart1...@gmail.com>wrote:
> Ah, hadn't used that before. And frankly I wouldn't bother doing that
> since it is a one-time operation. I would write a script to do the load
> with calls to psql (assuming you are using postgres of course). If order is
> not guaranteed then even if things work for you now using "initial sql" may
> just be postponing a problem to the future.
>
This is something we need to work from syncdb. And I am using MySQL.
> On 2013-12-14, at 6:02 AM, Larry Martell <larry.mart...@gmail.com> wrote:
>
> On Fri, Dec 13, 2013 at 8:08 PM, Tom Lockhart <tlockhart1...@gmail.com>wrote:
>
>>
>> On 2013-12-13, at 1:48 PM, Larry Martell <larry.mart...@gmail.com> wrote:
>>
>> I have 2 tables that I need to initialize with some initial sql files.
>> One table has a foreign key reference into the other. The docs say
>> "Note that if you have multiple SQL data files, there’s no guarantee
>> of the order in which they’re executed." So what can I do here? I
>> obviously need the table with the referenced data loaded first so I
>> can pick up the ids when I load the referencing table? How is this
>> situation (which seems like it would be fairly common) typlcally
>> handled?
>>
>>
>> If you already have the keys (that is, the table ids) assigned, then you
>> can probably load the tables in one transaction since integrity is checked
>> on commit.
>>
>> If you need the keys assigned, then you might want to do a python script.
>> psycopg2 can return a value which you can then stuff into your other table.
>>
>> But maybe your question has a simpler answer. If you are loading SQL data
>> directly, and not through django fixtures, you can control which table gets
>> loaded first. Load the one with the foreign key references second.
>>
>> You might need to set the next value for the sequences used for table ids
>> if you are doing a bulk load. Something like this for pgsql:
>>
>> select max(colname) from tablename;
>> select pg_catalog.setval('tablename_colname_seq', maxid, true);
>>
>> hth
>>
>>
> As I wrote in my initial post, I am loading the data with initial sql
> files (
> https://docs.djangoproject.com/en/1.5/howto/initial-data/#providing-initial-sql-data).
> I have to experiment and see what happens first - fixtures or initial sql.
> I may be able to do the referenced table with one and the referencing table
> with the other. Seems bulky though.
>
>
--
You received this message because you are subscribed to the Google Groups
"Django users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to django-users+unsubscr...@googlegroups.com.
To post to this group, send email to django-users@googlegroups.com.
Visit this group at http://groups.google.com/group/django-users.
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-users/CACwCsY7%2BWFLW_tkHCqzd74XUObfUQ%3DqExT%3D4MkfEz8qD6hTBsg%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.
