Hey Rocky,
Do you have any expected date for the final 3.5 framework? I know its Q1 2008. On the other hand do you plan to write book about the 3.5.
Technical Part:
I seen heavily use of reflection, which is so clean indeed, but is there any performance complaint so far?
Thanks
Dodi
Please see his blog entries:
http://www.lhotka.net/weblog/CSLANET35StatusUpdateDec5.aspxhttp://www.lhotka.net/weblog/CSLANETStatusUpdateNovember2007.aspx
And in terms of reflection, the general statement made in this area is that the ~15% performance loss (a number I think Rocky has used) is weighed against the practicalities and usefulness of the actual reflection.
Perhaps not an entirely thorough reply but hope that helps.
skagen00:
And in terms of reflection, the general statement made in this area is that the ~15% performance loss (a number I think Rocky has used) is weighed against the practicalities and usefulness of the actual reflection.
That 15% number was for a prototype reflection-based DAL that I wrote - and discarded. I thought that was too high a price to pay for data access.
The use of reflection in the data portal is incidental given the high cost of talking to a database and/or going to an app server. Essentially there's no impact in any real sense.
The use of reflection in n-level undo can have an impact on very large object graphs. But most people never see an issue, because this is used behind a smart client, so there's a dedicated client workstation running this code.
It is actually Q2 - probably June, maybe even July (as I might try to sneak in some ADO.NET EF and/or Silverlight content). (http://www.lhotka.net/cslanet/Roadmap.aspx)
The use of reflection isn't as bad as people seem to think. You almost certainly use it every day - if you use the BinaryFormatter, XmlSerializer, any form of data binding, WCF, WPF, WF, etc. They are full of reflection.
CSLA uses reflection in two primary places - n-level undo and the data portal.
Of those, n-level undo is by far the heaviest user. Of course n-level undo is an optional feature, and if you don't use it you incur no overhead. If you do use it, you are probably creating either a Windows Forms or WFP UI - in which case you are already neck-deep in reflection thanks to the way those technologies implement data binding.
On the data portal side the use of reflection is relatively indidental given the overhead of talking to the database and/or going across a network to an app server (http://www.lhotka.net/Article.aspx?area=4&id=4320e7f2-dd14-4928-b552-4eb0cb82cd68).
It is actually Q2 - probably June, maybe even July (as I might try to sneak in some ADO.NET EF and/or Silverlight content). (http://www.lhotka.net/cslanet/Roadmap.aspx)
Sorry, I thought you were asking about the books...
The framework is slated for a Q1 release, that's true. I expect to start putting out meaningful test versions toward the end of January, and if all goes well it should be "done" at the end of February.
I say "done", because the reality is that I change things as I write the books. That book writing process is really a quality check - not only that the code works, but that it is comprehensible and explainable. Every time I've done this the framework has changed during the book writing process, and I doubt this time will be different.
So 3.5.0 will be out, but by the time the book is done in June/July I'm sure the framework will be at some higher point release - .2 or .3 I suppose.
Copyright (c) Marimer LLC