Re: an attempt to make clsql more thread safe

On May 9, 2012, at 8:34 AM, Nathan Bird wrote:

> On 05/09/2012 03:55 AM, JTK wrote:
>> 
>> As discussed earlier, I had a stab at making clsql more thread safe.
> ……..
> Thanks a bunch for this work! My initial skim-through impression is that the code looks good but there is a *lot* of changes there.
> 
> Do you think you could try breaking it into some smaller pieces so we can go through this and review it a chunk at a time?  It's hard to digest a big hunk of changes like that and you're the best one for dividing it into smaller pieces. Having separate patches for the three items mentioned above would be a great start-- finer grained if it makes sense. ……….




Sorry about the big code dump; I wanted to get this out because other tasks were looming
and I wanted to make public something that at least passed the tests and seemed to work, rather
than having it sit in my  directory, forgotten.

I'm not very experienced with version control systems.  I should be able to git-ify this into the latest 
tree, but it might be a little while before I have a chance to do it.

The mysql fixes are fairly simple.

But the other parts are complicated because they are interlinked, and represent a pretty big
reworking of the functionality, so they're not really incremental.

Also, when working on this, I found database pools to be straightforward and elegant, but *connected-databases*
and find-database to be a bit byzantine.  It's as though there are two different ways of storing information
of open databases.     Is this intentional or  historical?  It seems to make sense to try
to get rid of *connected-databases* and give pools a bit more functionality (like having pools work in
non-pooling mode with :if-exists, etc).  



JK

_______________________________________________
CLSQL mailing list
[email protected]
http://lists.b9.com/cgi-bin/mailman/listinfo/clsql