<div dir="ltr"><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Feb 25, 2016 at 9:10 AM, Bengt Kleberg <span dir="ltr"><<a href="mailto:bengt.kleberg@ericsson.com" target="_blank">bengt.kleberg@ericsson.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">2) Two binary packages/libs that I need must have different versions
of some third lib. </blockquote></div><br>This isn't a namespace problem. </div><div class="gmail_extra"><br></div><div class="gmail_extra">It is a very interesting problem, though, and a difficult one. There are solutions elsewhere that one could learn from (for example, Java's OSGi), but like it was pointed out before, we have more than module names that are global and that should be compartmentalized: process registry is the most important one. Imagine if the third lib starts and registers a process - there should be two of them, and each lib instance should only know about its own process. </div><div class="gmail_extra"><br></div><div class="gmail_extra">Going back to the original question, introducing namespaces brings up similar issues: what should ?MODULE resolve to? We will probably need ?FULL_MODULE too, but when to use each? It's possible it will become necessary to convert between the short and full name of a module - how? Are macros still going to work or an introspection service is needed? </div><div class="gmail_extra"><br></div><div class="gmail_extra">I think these are definitely worth investigating, answering and solving.</div><div class="gmail_extra"><br></div><div class="gmail_extra">best regards,</div><div class="gmail_extra">Vlad</div><div class="gmail_extra"><br></div></div>