Magnus Hagander wrote:
> Attached patch implements wal_sync_method as an enum. I'd like someone
> to look it over before I apply it - I don't have the machines to test
> all codepaths (and some of it is hard to test - better to read it and
> make sure it's right).
> In order to implement the enum guc, I had to break out a new
> SYNC_METHOD_OPEN_DSYNC out of SYNC_METHOD_OPEN where it was previously
> overloaded. This is one of the parts where I'm slightly worried I have
> made a mistake.
I took a look at this and I like it -- it seems correct as far as I can
tell, and it has the nice property that we can display the list of
values that the current platform actually support.
This thing looked a bit dubious to me at first:
> ! switch (new_sync_method)
> ! case SYNC_METHOD_FSYNC:
> ! case SYNC_METHOD_FSYNC_WRITETHROUGH:
> ! case SYNC_METHOD_FDATASYNC:
because at least some of the values are only present on some platforms.
However, I think realized that the actual values are present on all
platforms, independent of whether they represent a supported fsync
method, so this is OK. Perhaps it would be good to add a comment on top
of the switch statement explaining this. Otherwise looks fine.
Well, and the ereport("FIXME") needs to be improved :-)
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.
In response to
pgsql-patches by date
|Next:||From: Andrew Dunstan||Date: 2008-04-07 22:28:07|
|Subject: Re: [PATCHES] Integer datetime by default|
|Previous:||From: Andrew Dunstan||Date: 2008-04-07 20:58:34|
|Subject: Re: Headers dependencies cleanup|