Hi all,

There's stuff happening on the wiki. I like that. I prefer to give my 
comments here though, as it's easier to keep track on the discussions.

== FileIO ==

On handling realtive paths. We need to decide between implicit or 
explicit management. I proposed implicit first (storing curren path for 
each protocol & thread) but have lately moved to explicit (loaders are 
responsible for mapping that) as I think the first is too error prone 
and not really worth the effort (it is sexy, I'll admit to that, but how 
much maintenance will it require? ;).

I've also put in some sample code there, to try to clarify some concepts.

Comments?

== ExamplesDemosTutorial ==

We seem to have agreed on moving Examples out of the Source-tree? I like 
that.

I also agree with Dirk on having demos _outside_ of the dailybuilds, but 
maybe in a 'Point Release', I.e. something like the large ATI/Nvidia 
SDKs. (OpenSG-2.X-SDK.. yeah.. sort of like java & .NET & everything 
else.. :) Reasons: Demos should be big and bulky with fancy content.

Minor nitpick: I don't thing tutorials need to be numbered. A doxy-page 
listing them in some categories as (beginnner/intermediate/advanced) 
would probably suffice.

There doesn't seem to be much more to discuss here? or?

== Pointer ==

Could someone (Gerrit? you seem to be 'in control' ;) clean up the wiki 
site summarizing current 'consensus' and points of discussion? That type 
of discussion would probably work out better on the mail-list as we can 
follow threads and comment more easily on things.'

My (IMHO) comments on the current pointer discussions:

- I suggest using boost::intrusive_ptr as much as possible, for our 
internal ref counting, if it is applicable. It eases interoperation with 
other boost stuff, unless someone is willing to work to make it 
compatible. :)

- I'm fine with passing out c-ptrs & create() not calling addref. 
(That's how it should be). However, create() ought to return a RefPtr 
for safety. I strongly feel that public API calls should not leak more 
than necessary, and create() is such a common things so.

- I like the idea of a 'mapping pointer' similar to fcptr today, to use 
for safety. To have that as a refptr as well would probably also be 
good. Provides a 'totally safe ptr' for combined safety and comfort. 
(intializable from c-ptr so I can do Node* n  = foo::create(); 
NodeMegaPtr p(n);).

- For MPtrFields iterators there's boost::indirect_iterator which might 
help out?

- I would really like to avoid calls to addref/subref in any OpenSG 
code, if at all possible (that's not really up to me though). Our 
framework here manages with boost::intrusive_ptr, our own Refobject 
class and weakptr or c-ptr's for cycles/"back-pointers".

== Documentation ==

Is it possible to run some kind of diagnostic (ala boost does for 
licenses etc) to see if classes/methods are documented or not? That way 
it would be easy to monitor the doc process?

/Marcus


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to