On Thu, Oct 26, 2017 at 2:15 PM, Clément Bera <bera.clem...@gmail.com> wrote:
> Hi, > > Good work, > > In PolyMaths/SciSmalltalk vectors and matrix are already implemented > efficiently. I would recommend to share the same implementation. > By this, you mean move the "base" vector/matrix objects from there into Pharo? This makes a lot of sense. These are applicable to a lot of applications, and having them in-Image obviously helps focus efforts and reuse. These could just have the minimal methods required by Bloc, with PolyMaths/SciSmalltalk extending them with more exotic operations. > In the next few years, I think we will improve the performance of those > vectors/matrix library using the processor vector instructions, and it > would be nice if we have to improve one but not two libraries of vectors > and matrix (This is the polite way of saying I am going to improve one > library and the others will have to do it themselves). > > Regards, > > On Wed, Oct 25, 2017 at 5:53 PM, Tudor Girba <tu...@tudorgirba.com> wrote: > >> Hi, >> >> Our team is working on a couple of projects that are relevant for the >> core of Pharo, namely GT and Bloc/Brick. At ESUG we were asked by several >> people, and Stef in particular, to make the progress on these projects more >> transparent. To this end, we will start two streams of signals: >> - fine grained info bits on out Twitter account (several times a week): >> https://twitter.com/feenkcom >> - a less often but regular (probably every 2 weeks) activity log message >> sent to this mailing list >> >> Please let us know if you see an issue with this. >> > Good initiative, and already getting to useful collaboration. btw, I know sharing opens up to nit picking that can be discouraging, as we've seen other tools you've introduced in the past get a bit of a beating at the transition stage - but your vision and good work has borne through and now these are tools we now can't live without! So keep up the good work. cheers -ben > >> In the meantime, let’s start. >> >> Bloc: >> - Over the past couple of weeks Alex worked on transformations and >> measurements in the core system. It turns out that there was room for quite >> a number of performance optimizations and for making the system more >> debuggable. >> - At the low level, this involved adding matrix and vector support. >> https://twitter.com/feenkcom/status/923123870537863168 >> https://twitter.com/feenkcom/status/916300180848349185 >> - At a higher level, it implies keeping track of measurements >> https://twitter.com/AliakseiSyrel/status/915203525931622400 >> - And of transformations: >> https://twitter.com/feenkcom/status/919603739656417281 >> https://twitter.com/feenkcom/status/917116164484096001 >> - The BlScalableElement is now working properly >> >> GT Examples: >> - The current repository with unary-based examples is at: >> https://github.com/feenkcom/gtoolkit-examples >> - We now have a runner in the inspector that allows us to both see the >> source code and run and play with the resulting object: >> https://twitter.com/feenkcom/status/923210686204989442 >> - Andrei made the “after” behavior to work with unary examples. >> >> Cheers, >> The feenk team >> >> -- >> www.tudorgirba.com >> www.feenk.com >> >> "Being happy is a matter of choice." >> > +1