Magma question

Chris Muller chris at funkyobjects.org
Sat Nov 4 23:23:57 UTC 2006


Hey Göran, congratulations Gjaller.  I saw your presentation (the BOF video) and it looks very impressive!  I think we should push (eventaully, at least) to use Gjaller to manage the Squeak development process and bugs.

> The more interesting fact is that Magma indexes seem to do their job
> perfectly fine with 600 cases in Gjallar.

The indexes have been very heavily tested, I don't expect you will have any problems with 10000, 100K or even much more (other than, perhaps, the time to load them!).

As for the buffer reading, is it possible perhaps that you were asking for the #root in there somewhere?  That allways forces a buffer read.  If you were only sending #at: to a MagmaCollectionReader, then I agree it doesn't seem like it should be doing any buffer reads..

I'll do some more thinking and poking around, or if you have more detail about what operations you were doing during the profile that would probably help.

Regards,
  Chris

----- Original Message ----
From: "goran at krampe.se" <goran at krampe.se>
To: afunkyobject at yahoo.com; magma at lists.squeakfoundation.org
Sent: Friday, November 3, 2006 8:05:22 AM
Subject: Magma question

Hi Chris!

Ok, so I am frantically getting Gjallar into some deployable shape here.
:) Yesterday I played with creating 600 cases programmatically - kinda
complex objects, one per commit - took 25 min which isn't all bad - and
it was doing other silly things too.

The more interesting fact is that Magma indexes seem to do their job
perfectly fine with 600 cases in Gjallar. One of these nights I will
push 10000 into it and we can see what happens.

BUT... one silly question:

If I use a reader with an index query - and the cases that the reader
finds are already cached into the session - should it really be doing
buffer reading? While doing one of these operations using the profiler I
kept hitting "Q2Case allInstances size" in a workspace and it constantly
showed 85 - just like before. So no new Q2Case objects were created but
the profiler spent 10 seconds in Magma doing buffer reads it seems. I
had expected it to just get some kind of oid, discover it already has
the object in the imag - and that would be it.

I can of course provide more info if you need it.

regards, Göran
_______________________________________________
Magma mailing list
Magma at lists.squeakfoundation.org
http://lists.squeakfoundation.org/mailman/listinfo/magma






More information about the Magma mailing list