[Uni-verse] Purple State of the Art report

[Emil Brink]

Lauri Savioja wrote:
> Hi Emil,

Hey Lauri! A good sign that everything is back to normal after the
vacations, your quick replies are here again, heh.

> I had a quick look, and for me the document looks nice. I have one
> proposal related to Introduction. In my opinion you could add a
> sentence or two of the aim of Purple. Now you go directly to details
> without even mentioning Verse and its relation to Purple.
> 
> If we give this document to some imaginary nasty reviewer, he might
> find some other issues as well. I mean that by reading your text one
> could get a feeling that there are two main reasons for implementing
> Purple:
> 
> 1. You wanted to implement it
> 
> 2. Verse is too incompatible with any other approach such that the
> most easy way was to start from scratch
> 
> Both of the reasons are quite poor. I don't know how to soften this
> issue, but you could still give a second thought to what is the real
> message you want to convey to a reader. Please note: In my opinion
> Purple is very good idea and personally I don't have anything against
> it.

Right. Thanks. I understand that's a valid point, partly probably
caused by myself being a bit "blinded" by it from spending all that
time on it.

I guess the reason *I* did it was "it is in the spec, so someone had
to do it". It seemed rather interesting, and I didn't object much, but
I also didn't come up with the idea, never fought for a chance to do it, 
and so on. I'm not that prepared to defend it.

Eskil, care to comment here? I should add some explanations to the 
report, but would very much like your input.

Regarding the second reason, I'm not so sure that it's all that bad
if that were the case, really ... I guess it depends on how you look
at things. If Verse was very similiar to some existing system, the
entire project would fall for the same argument. If it isn't, then it
makes sense to have a "native" data flow environment too, I think.

> The comparsion with OpenDX is OK, but the statement of pd is a bit
> superficial. Now you give an impression that you didn't want to look
> how good it would be. It would help, if you change the wording from
> "OpenDX is likely the better fit" to something more convincing.

Yes! That was never my intention. I guess the core reason is that I
don't *know* Pd (or OpenDX, for that matter). I haven't been a good
enough scholar to spend all that much time trying to familiarize myself
with them either, to be honest. I find that answering questions about
internal data paths and such from documentation is very hard, but going
the other way and learning these big systems has felt like too big an
undertaking.

> One typo: "... inf ..." -> "... if ..." in the last paragraph of your
> text.

Okay, fixed. Thanks.

> Emil and Eskil, could you please send me some existing presentation
> material of Verse. I'll have a talk of Verse and Uni-Verse in Finnish
> SIGGRAPH seminar early September, and I think the result would be best
> if I could use your slides :-).

I've never done any formal presentation of Verse; I don't have any such
slides. I'm not sure what Eskil might have, since he seems to prefer a
rather ... informal presentation style. I do know that Gert has at least
some material, so perhaps he can help you out?

Regards,

/Emil