On 07/31/2014 05:58 PM, Simon Riggs wrote:
> On 31 July 2014 15:48, Heikki Linnakangas <hlinnaka@iki.fi> wrote:
>> On 07/31/2014 05:19 PM, Simon Riggs wrote:
>>>
>>> On 31 July 2014 14:59, Heikki Linnakangas <heikki.linnakangas@iki.fi>
>>> wrote:
>>>
>>>> Move log_newpage and log_newpage_buffer to xlog.c.
>>>
>>> Got a feeling this wasn't properly discussed because its unlikely
>>> anybody would agree with adding more stuff to xlog.c
>>
>> Well, they certainly don't belong in heapam.c... Got a better suggestion?
>
> I guess if we had xlog_fpi.c with RestoreBackupBlock et all, plus the
> stuff you just moved it might make sense.
>
> Just looking for ways to prune it down.
Yeah, something like that might make sense. I totally agree that xlog.c
is huge (although so is heapam.c). In the WAL format & API patch, I'm
putting much of the new code for constructing a WAL record into a new
file, xlogconstruct.c (in the latest version I'm just about to post -
the previous version put them all in xlog.c).
>> (I bumped into this again while working on the WAL format & API changes,
>> which understandably rewrites those functions again. It seemed better to do
>> this refactoring separately than include it in the huge patch.)
>
> Yeh, no problem with the refactoring itself, but I wanted to remind
> you to discuss stuff.
Ok, thanks. I did post to pgsql-hackers first, see
http://www.postgresql.org/message-id/53DA0ECF.4040002@vmware.com.
Admittedly I didn't leave much time for discussion between the post and
commit; that was because I felt quite certain that people wouldn't
object to this.
- Heikki