<div dir="ltr"><div><div><div>2014-04-01 14:22 GMT+02:00 "Jóhann B. Guđmundsson" <<a href="mailto:johannbg@gmail.com">johannbg@gmail.com</a>>:<br>><br>> On 04/01/2014 11:57 AM, Christian Schaller wrote:<br>
>><br>>> Well I think the real disagreement here is what the products is supposed to be<br>>> about. The WG wants to create an integrated product that is clearly identifiable,<br>>> not a 'mix your own' cookbook. We want to the Workstation to be as clearlyidentifiable<br>
>> and targetable a product as MacOSX is, while I think you want the workstation to be the<br>>> equivalent of shipping a random desktop system on top of Darwin and still<br>>> claim it is 'MacOSX'.<br>
>><br>>> Christian<br>><br>><br>> Which is equivalent of saying that the serverWG should tie them to a single<br>> server role.<br><br>I don't think this is quite precise. The Server WG is, essentially, inventing a new, <i>single</i>, D-Bus API for roles (and assuming a <i>single</i> glibc API for server implementation[1]) . On the desktop, we already have GTK[23], Qt[45], and more fundamentally disjoint alternatives, and from time to time <i>actually matters</i> whether applications are all using a single one, or many (e.g. new features have to be implemented once vs. ~four times).<br>
</div> Mirek<br><br></div>[1] Well, and it gets really complicated above glibc. The above-glibc layer is really fragmented and there isn't much consensus over whether any particular library is "canonical" / " official". "Luckily", many server implementations want to be portable to different OSes so they tend to reimplement these libraries instead using them, so the fragmentation doesn't matter... which in effect makes the fragmentation even worse.<br>
</div> Mirek<div><div><div><br></div></div></div></div>