On Mar 24, 2008, at 8:28 AM, devzero2000 wrote:

> rpm 5, from version at least 4.4.6 seems, seem to offer all new  
> types of dependencies resolution, but there is not or little  
> information about it: The notes in the changelog do not illustrate  
> at full the potentialities or at least the functionality. For  
> example dependency name space, macro expanded dependency tokens,  
> boolean valued  dependency functions.... ( iirc  from a mail of  
> some year ago )
>

There's little usage of probe dependencies either.

Typically it takes 2-3 years for rpm implementations to be adopted.

Which means that I have another year or two to document
probe dependencies ;-) Seriously, I don't believe lack (or
existence) of documentation will affect rate of adoption at all. YMMV.

Meanwhile, each probe dependency was described when I
implemented on the old <rpm-devel@lists.dulug.duke.edu>
in some detail. I'm sure you have found the information already.

There have also been some efforts at http://rpm5.info to describe
runtime probe dependencies. But without real world usage cases,
its rather an academic exercise.

> Certainly, also the order of the dependencies is changed from the  
> legacy rpm (think to erasure ordering)
>

Yup. all depends on what meaning one wants to attach to "erasure  
ordering".
What is in rpm-4.4.2.3.4.5.6 is known feeble and broken. Gud enuf seems
adequate for those without clue.

> It is possible to have some information in more on these topics?  
> Someone use it already ?
> To me they seem too much interesting.
>

Meanwhile feel free to ask whatever questions you have. I can
easily (and happily) supply information as needed. If there's
sufficient need, I'll write up what is currently implemented.

Again, I do not believe lack/existence of documentation has
any relation to rate of adoption. The release engineering to
deploy a new version of rpm iis non-tivial, and s what limits
the rate of adoption of Newer! Better! Bestest! features in *.rpm
packages.

73 de Jeff