Re: [Monotone-devel] Re: Windows version 0.36

[J Decker]

frigging mailing lists responding to wrong places

 

<unquote>

 

On 8/31/07, J Decker <address@hidden> wrote:

I happen to now know a bit about what sqlite does with locking. 

If you just do

 

prepare( "select blah from somewhere" );

step

 

and this does not either get unmade - err destroyed?  I forget exactly what the close is

 

or if't snot at the end of the result set..

 

followed by an insert

 

creates an externally locked condition.

 

 

"Clean up your freaking queries!"

 

 

</unquote>

On 8/30/07, Derek Scherger <address@hidden > wrote:

J Decker wrote:
> This is only new as of 0.36... previously any locks were unlocked...
> this is more like a sign that some query is left hanging instead of
> being completed.  It's not that I'm attempting to use it at the same
> time, but, again previously, I could push/pull from the database and
> then check out with the server process still running.  (Not that it was
> serving anything at the same time which definately resulted in
> a database locked condition)

I don't know of anything that changed regarding database locks in the
0.35 - 0.36 timeframe but I haven't been following all that closely. If
this did work previously it might just have been luck, it's not designed
to as far as I know. i.e. I don't *think* there's any code that says
"close the database after this netsync connection is finished with it".

I've replied to the list as well in case someone else knows differently
though.

Cheers,
Derek