RE: 4.0.5 ?

[Borsenkow Andrej <Andrej.Borsenkow@mow.siemens.ru>]

> >
> > IMHO there can be no 4.2 unless module dependency problem is fixed
> > (dynamic modules that depend on other dynamic modules).
> 
> From the perspective that the current dev branch is almost as stable
> as 4.0 it would make some sense to declare a 4.2 to allow the users
> to benefit from those 229 changes. 

I am not sure if they necessarily benefit from them. I did not analyze
every article but at least some changes are related to code rearranging;
I guess most real bug fixes and new completion functions did go into
4.0.


> If we hold out for big changes,
> they may not happen for a while and then they'll take a while to
> stabilise. I don't think the long 3.0 - 4.0 gap was ideal.
> 

Then please put current state of affairs into BUGS. It is bug after all.

> > To remind - it is impossible to dlopen() a shared object that has
> > unresolved data objects even on Linux, that currently rules out any
> > solution that encodes dependencies in binary itself.
> 
> Would it work on platforms like Linux to link modules against those
> they depend. Using external files would be ugly but I'd mind less if
> they were only needed on some platforms.
> 

I am not sure what do you mean. Several modules in one shared object?
This requires even more changes.

I see that nobody (me including) is happy with extra files and nobody is
going to undertake major rewrite :-) I do not have much spare time
currently, but I still hope to find some automatic way to replace all
variables with functions.

-andrej