> But this gives me an idea - the pointer object could perfectly well
> simply cache the previous item in the list so that the current one can be
> deleted and the previous one linked to the successor.

I will try to implement that! anyway, I think this is rather an implementation 
detail...

> (Of course if the window's visible it's still very inefficient - so this
> might be a lot less important than I've been thinking it is.)

I'd say if efficiency is of uttermost importance, one shouldn't work with large 
lists of scalars in the first place :-)

Christof

> Gesendet: Dienstag, 04. September 2018 um 06:00 Uhr
> Von: "Miller Puckette" <[email protected]>
> An: "Christof Ressi" <[email protected]>
> Cc: "[email protected]" <[email protected]>
> Betreff: Re: Re: [PD-dev] roadmap for Pd-0.49?
>
> I'd like to find a way that doesn't use glist_delete() which does a
> linear search to find the objects's predecessor - if you use this to cull
> items from a list of them the complexity is order n^2.
> 
> But this gives me an idea - the pointer object could perfectly well
> simply cache the previous item in the list so that the current one can be
> deleted and the previous one linked to the successor.
> 
> (Of course if the window's visible it's still very inefficient - so this
> might be a lot less important than I've been thinking it is.)
> 
> cheers
> miller
> 
> 
> On Sun, Sep 02, 2018 at 11:33:49PM +0200, Christof Ressi wrote:
> > Hi Miller, what problem do you see with this solution in particular: 
> > https://github.com/pure-data/pure-data/pull/383 ?
> > 
> > you're right that deleting a scalar by pointer will invalidate all other 
> > pointers in the glist, but there's really no way around it...*) it also 
> > happens when you delete a scalar with the mouse. if it's fine there, why 
> > not also allow it via a method? at least I'm providing a valid pointer to 
> > the next scalar in the list.
> > 
> > Christof
> > 
> > *) unless we change the whole ref counting system to be per scalar and not 
> > per glist - which I'm not going to advocate here :-)
> > 
> > > Gesendet: Sonntag, 02. September 2018 um 23:00 Uhr
> > > Von: "Miller Puckette" <[email protected]>
> > > An: "Henri Augusto Bisognini" <[email protected]>
> > > Cc: "[email protected]" <[email protected]>
> > > Betreff: Re: [PD-dev] roadmap for Pd-0.49?
> > >
> > > I hope to get the intelligent patching features in - but not sure about 
> > > the
> > > pointer-invalid problem; I've been thinking for years about how to do that
> > > and haven't liked any of the solutions I've seen yet.
> > > 
> > > cheers
> > > Miller
> > > 
> > > On Sun, Sep 02, 2018 at 02:52:37AM +0000, Henri Augusto Bisognini wrote:
> > > > Hi! Since we got on the subject of 0.49 do you guys mind if i ask a 
> > > > couple questions about what's planned? Just out of curiosity and 
> > > > excitement : )
> > > > 
> > > > -I see there's a PR (#383) that adds a delete method for [pointer]. 
> > > > There's any forecast for that feature? I've been waiting that one heh
> > > > 
> > > > -Also will 0.49 come with the intelligent patching features IOhannes 
> > > > been working on? That will be a dream come true.
> > > > 
> > > > Cheers
> > > > Henri.
> > > > 
> > > > 
> > > > ________________________________
> > > > De: Pd-dev <[email protected]> em nome de Miller Puckette 
> > > > <[email protected]>
> > > > Enviado: s?bado, 1 de setembro de 2018 21:28
> > > > Para: Christof Ressi
> > > > Cc: pd-dev
> > > > Assunto: Re: [PD-dev] roadmap for Pd-0.49?
> > > > 
> > > > Looking back I've only been able to get out a bit fewer than one major
> > > > release per year... I hope to speed this up now that there are much 
> > > > better
> > > > scripts and tools in place (thanks mostly to other people!)
> > > > 
> > > > My own situation is that I go into heavy teaching mode again Sept. 24 so
> > > > I'll either succeed and get out 0.49 before then or fail and only get 
> > > > it out
> > > > late December.  If I do get 0.49 working quickly I'll then aim for 0.50 
> > > > in
> > > > February or March.
> > > > 
> > > > For 0.49 I want to fix the escaping of '\' characters (I had earlier 
> > > > tried
> > > > to make it possible simply never to make a `\` character visible to the
> > > > user but this has turned out to be impossible).  Also I want to fix 
> > > > pasting
> > > > to place objects under the mouse (or at the very least, somewhere 
> > > > visible,
> > > > not off the edge of the window)
> > > > 
> > > > For the medium term I'm thinking about revisiting array drawing to make 
> > > > it
> > > > more efficient and controllable, and expanding the capabilities of the 
> > > > 'atom'
> > > > box (lists; texts).
> > > > 
> > > > Also I have a vague plan to make an API feature to guide DSP graph 
> > > > sorting
> > > > order, e.g., to request that one item in a patch be sorted in advance of
> > > > another without them being connected by a patch cord.
> > > > 
> > > > Anyway, I think whatever happens to 0.49 I can deal with feature PRs 
> > > > anytime
> > > > this fall and get them into a release by late winter.
> > > > 
> > > > cheers
> > > > Miller
> > > > 
> > > > On Sun, Sep 02, 2018 at 12:32:47AM +0200, Christof Ressi wrote:
> > > > > Hi Miller,
> > > > >
> > > > > I just wanted to ask what's your approximate roadmap for Pd 0.49? Do 
> > > > > I need to hurry up if I want to propose stuff or is there plenty of 
> > > > > time left? :-)
> > > > >
> > > > > Christof
> > > > 
> > > > _______________________________________________
> > > > Pd-dev mailing list
> > > > [email protected]
> > > > https://lists.puredata.info/listinfo/pd-dev
> > > 
> > > > _______________________________________________
> > > > Pd-dev mailing list
> > > > [email protected]
> > > > https://lists.puredata.info/listinfo/pd-dev
> > > 
> > > 
> > > _______________________________________________
> > > Pd-dev mailing list
> > > [email protected]
> > > https://lists.puredata.info/listinfo/pd-dev
> > > 
> 


_______________________________________________
Pd-dev mailing list
[email protected]
https://lists.puredata.info/listinfo/pd-dev

Reply via email to