[Devel] Fwd: Dynamic manifest rev. 2

[Stefano D'Angelo]

Sorry, forgot to click "reply all".

---------- Forwarded message ----------
From: Stefano D'Angelo <zanga.mail at gmail.com>
Date: 2011/5/16
Subject: Re: [Devel] Dynamic manifest rev. 2
To: David Robillard <d at drobilla.net>


2011/5/16 David Robillard <d at drobilla.net>:
> On Sat, 2011-05-07 at 11:10 +0200, Stefano D'Angelo wrote:
>> 2011/5/2 Stefano D'Angelo <zanga.mail at gmail.com>:
>> > Hi all,
>> >
>> > I wanted to write this mail before, but because of the mailing list
>> > crash, I chose to wait a little longer in the hope that interested
>> > people has subscribed back again.
>> >
>> > Straight to the point: I am planning for a revision 1.2 of the dynamic
>> > manifest extension. The changes I'm confident about would be:
>>
>> I mean rev. 2
>>
>> > * the extension URI gets finally changed to
>> > http://lv2plug.in/ns/ext/dyn-manifest/ so that it follows the same
>> > naming pattern as the others;
>> > * dman:DynManifest becomes a subclass of lv2:Resource, thus
>> > dman:optionalFeature and dman:requiredFeature are substituted with
>> > lv2: equivalents;
>> > * all functions will belong to the discovery category as far as
>> > threading rules are concerned (thus removing all of that mess I wrote
>> > down last time in this regard);
>> > * some clarification in the header documentation (e.g., what is
>> > subjects, what is data).
>
> Fine.
>
>> Yet another idea: the dynamic manifest plugin library must not specify
>> lv2:binary, which may be instead implicitly be added to the data by
>> the host. This would allow to change the name of the shared library
>> without requiring rebuilding (only changing the manifest).
>
> I don't understand this one.

If doing so, you don't need to rebuild the binary library if you just
want to change its filename, just editing manifest.ttl would be enough
(of course, unless the lv2:binary is another one - in that case you
would be allowed to specify it, that is hosts should add or imply
lv2:binary if none is given).

>> > I don't think that this is going to break anything in practice, even
>> > though technically speaking we are breaking the API (?)...
>
> Are we?

Changing spec URI + changing RDF types sounds like that, no?

>> > If you have any problem with it, speak now or forever hold your peace. :-)
>>
>> You have time until Monday morning CET.
>
> Extensions in that namespace do not get released until there are at
> least two independent working implementations, you don't get to make
> such ultimatums... especially when you havn't shown us the actual
> extension ;)

My effort on lilv and zynjacku was implied ;-)

Stefano