[Date Prev][Date Next] [Chronological] [Thread] [Top]



> > - The basic design will be similar to that of Netscapes Java-SDK.
> You know the internet drafts regarding LDAP under
> http://www.ietf.org/ids.by.wg/ldapext.html
> especially
> draft-ietf-ldapext-ldap-java-api-11.txt and
> draft-ietf-ldapext-ldap-java-api-asynch-ext-05.txt?
Yes, I know them. Netscapes Java-SDK implement these drafts.

> first glance. Other problems might show up, while you're developing
> your class hierarchy for C++. But it will surely serve as a good start
> and discussion point.
Thats exactly what I thought :)

> Yes, I'd do so, too. Both the STL and exceptions are very useful. But
> expect some problems even with modern compilers. It's a pity that
> these problems still exist.
Is this really still a problem? I thought this problems were solved, since 
standard is already a few years old. Do you know what compilers could make 
problems? For the implementation I will use the GNU C++  Compiler (2.95.2 or 
newer) I don't expect problems with that one. How is it with the MS or Borland

> Regarding the STL you could of course avoid it in your specification
> and make this an implementation specific detail.
Yes thats right. Shouldn't be a problem

[..] (deleted some stuff I totally agree with you :) )
> > - I plan to use some parts of the C-API inside of my C++ API (e.g. the
> >   Network I/O and BER-encoding/decoding). I think this would reduce possible 
> >   bugs (and work :) ) a lot. The thing is that I didn't find docs for the
> >   network I/O stuff. Is there any, or do I have to look into the sources?
> What exactly do you mean by network I/O stuff? 
I was talking about the part of the OpenLDAP-C-lib that does the network
communication.  AFAIK it's platform independend (runs on UNIX Systems as wells 
on WIN NT) and I think it good to use it for the Network I/O. The drawback
is that it is not really documented. So I have to look into the libldap-sources 
to figure out how it works. 

> I think it's not a
> problem to start wrapping the C-API calls. One could still extend this
> by "pure" C++ functions and methods later if problems of any kind
> arise.
> For a diploma thesis it could of course be a good idea to dig
> deeper, just for the effects of learning. But this is up to you, of
> course. And the C-API calls are documented quite good, I think.
Just what I plan to do :)
Thanks for your comments,