September 2017 - Pharo-users

Smalltalk gets a reference in fastcompany Kay interview

by Tim Mackinnon

Interesting (although I found slightly rambling) interview with Alan Kay about personal computing. But Smalltalk does get a link. https://medium.com/fast-company/the-father-of-mobile-computing-is-not-impre… Tim Sent from my iPhone

Sept. 17, 2017

Re: [Pharo-users] Pharo 7 license question

by Dimitris Chloupis

Mr lawyer here I will give you general direction because in the end it depends in the national law of the country of the person being sued. The general idea is that GPL is a license to be avoided if you want real freedom for your users. That means their ability to open or close code. A license is in essense a contract that you accept when you click a button, a link or other means that show clear consent. If the license is not accepted through the means provided by law then it cannot apply. Also a license is bound by law and national laws have special legislation of what contracts can have as terms. Anything that may cause bodily harm for example or negate the liability for such action are terms in the a contract that are immediately become invalid and hence they dont apply without necessarily make the whole contract invalid. Basically means that a license by itself is powerless and can function only under the permissions of law. A GPL license is thus a contract, but a GPL license as we all know it, generally is used as a template. None forces anyone from making a GPL license that allows the the user to close the code. FSF could sue for the use of the name on the ground of misleading the user, but the license would be still valid. Matter complicate also on the matter of GPL "infection". One licences his code under MIT but uses GPL licensed libraries, keeping his code open does not violate GPL but if another user take the code and add his own code to it and close it , including the GPL license he does violate the GPL license but does he violate the law ? My legal opinion on this is that he does not unless there was clear consent by him on the matter of GPL license, which would require for him to know that those libraries are used by the MIT code he is using and those libraries are GPL. Taking into account how messy code bases can be, good luck with that. Again this depends on case and national law , hence why we say always , go speak with a lawyer of your juridiction , preferably someone who specialises in contract law. NOT Copyright law. A license has little to do with ownership , which is what copyright is, its merely a contract trying to bind you to use the code only in specific ways. Violation of a license is not a violation of copyright. Again under the exclusion of national laws, national laws have each their own weirdnesses. Also another advice, it does not matter if the license is called MIT, GPL, BSD , BOOGIE , or BIGPINKELEPHANT , the bottom line is that licenses are contracts and as such as soon as you accept them they are biding to the extend the law allows them to be binding , as I mentioned earlier Thus.... When we venture into the commercial world and profit is involved, people who will want a piece of that profit may emerge and use the license against you. Thus do not accept any license even if it is certified MIT before giving the text to a lawyer to read it and find any traps , hidden doors or anything that can put you in a compromising position. There are even cases when both parties have good intentions yet they entangle in a legal battle because the license is too vague, confusing or it does not specify a special case that may apply. Contracts can be extremely complex entities, even though I am not a copyright lawyer , I do general practice and martitime law, that Greek Law and European Law by the way, I know how hairy things can get , especially with big contracts. I also have a single experience with Copyright of the code , between an individual and the company and it was very very messy for both parties. Unfortunately judges still do not understand computers and I can tell you that also apples to other European countries and the USA because I have a master degreen on commercial and e-commerce from a UK uinversity with a disertation "Legal protections against malicious software". I examined UK, Eurpean and International law for this dissertation and what I have seen has severely disappointed me how law professionals, not just judges are so clueless when it comes to technology and yet are expected to provide fair protection and judgements. Summary : If you plan to make profit, go talk with your lawyer. If you intend to make a company a lawyer and an accountant is a MUST HAVE and wil save you tons of money than they will cost you. Also never wait for the last second to seek legal advise , in law everything is under a time limit, you dont want to be late. Trust me ;) On Sun, Sep 17, 2017 at 8:01 AM Jimmie Houchin <jlhouchin(a)gmail.com> wrote: > I understand that Pharo people will in general want to stay away from > the GPL. I just didn't know if it would potentially be more equivalent > to how other languages work. > > In Python to my understanding I could do something like > > #into my MIT licensed app > import GPL_library > import MIT_library > > And the GPL not be viral in my app provided I only use the GPL library > and am not modifying it in my app. > > Do I understand this wrong? > > I haven't yet investigated Pharo 7 yet, so I do not know what would be > similar to the above. > > MyMITClass #initialize > "Install configurationOfGPL package" > self useGPLPackage > > Yes it is contrived. But I am trying to see why Pharo would be different? > > Both are loading GPL code and running it in a mixed system. > > Regarding dual-licensing. I don't think that is likely. Most of these > people prefer the GPL and use it because they want its viral nature and > dislike the permissiveness of MIT software. They want to constrain they > software. > > Thanks to all who replied. I just thought I would ask. > > Jimmie > > > > On 09/16/2017 02:17 AM, Stephane Ducasse wrote: > > I do not think that the bootstrap changed anything. :) > > We will stay away from GPL. > > May be you can talk to the people of the libraries you want to use and > > see if they are interested in a dual license. > > > > Stef > > > > On Fri, Sep 15, 2017 at 10:49 PM, Jimmie Houchin <jlhouchin(a)gmail.com> > wrote: > >> Hello, > >> > >> Pharo 7 to my understanding fundamentally changes Pharo. It is my > >> understanding that Pharo 7 starts with a core Pharo kernel and like many > >> languages out there, imports or adds code from a variety of external > sources > >> to the image being built. > >> > >> With that understanding, I am curious if that would allow for inclusion > of a > >> specific library/module to be licensed as GPL? And it not affect the > other > >> code in the composed image? > >> > >> I am a big believer in the MIT/BSD license and not a big fan of the GPL. > >> However, there is software out there that I have avoided looking at the > >> source code or attempting to port it to Pharo because it is GPL. I would > >> sincerely love if I could now port such a library and license it under > the > >> GPL as required, and it not affect any other code outside of that > specific > >> library. > >> > >> I am not a lawyer. Nor do I know any lawyers. Is is possible for > someone to > >> get a reasonably definitive answer on this question? > >> > >> I am sure I am not the only one who has had this desire. I am also sure > that > >> I am not the only one who will have this question in the future. So it > would > >> be nice to have a proper legal response that could possibly be > explicitly > >> stated somewhere on the website or on an FAQ or something. > >> > >> Regardless of the answer, yes or no. It does need to be a settled issue > for > >> Pharo. That way someone could know if GPL/LGPL or whatever software > could be > >> in the catalog. > >> > >> Just wanted to put that out there to the community. I look forward to > the > >> answer, should one be or become available. > >> > >> Thanks. > >> > >> > >> Jimmie > >> > >> > >> > > >

Sept. 17, 2017

Re: [Pharo-users] Standalone HTML Model

by Norbert Hartl

> Am 17.09.2017 um 13:34 schrieb Stephane Ducasse <stepharo.self(a)gmail.com>: > > Norbert can you explain the difference between "a HTML model that can > be used for streaming, too, instead just the streaming." > because indeed in Pillar we have another HTML exporter and I would > like to reuse :) The difference is having a streaming API like seaside that uses brushes to write a HTML tag with its attributes and uses blocks to nest tags. This way at the end of the code you don't have a document model but a serialized form of the DOM that depends on the canvas you use. The standalone HTML model would be a DOM on which you can work afterwards. You can navigate, decorate etc. before serializing this the a html string. For the PharoJS this would ease the development in the pharo image. For pillar it would be a theoretical improvement to convert a pillar DOM to a HTML DOM and this to a HTML string representation. But as long as the benefits aren't clear the intermediate step for pillar likes a bit too much. But I have the strong feeling it would be good if there would be more directions you can go. We have a good XML parser that produces a XML DOM. We have pillar with its own DOM that can be converted to HTML. I don't know what SOUP is producing. And we have web frameworks where people have different ways producing for the web. I see potential in supporting multiple ways of reading and writing HTML and making this experience much more powerful. Norbert >> On Sun, Sep 17, 2017 at 1:29 PM, Norbert Hartl <norbert(a)hartl.name> wrote: >> >> >>>> Am 17.09.2017 um 12:12 schrieb Pierce Ng <pierce(a)samadhiweb.com>: >>>> >>>> On Mon, Sep 11, 2017 at 06:25:55PM +0200, Stephane Ducasse wrote: >>>> Hi Pierce Ng >>>> How different is the API from Seaside? >>>> Because I would like to use it. >>>> I like to think modularly :) >>> >>> Hi Stef, >>> >>> I modeled it after Seaside's API. However, it was really just a start and needs >>> much more work. I created this when investigating writing web apps in Cuis, >>> hence the perceived need for a standalone library. Eventually I ran out of time >>> and decided to stick to full Seaside loaded into Pharo. >>> >> So you've built something like seaside canvas? I like to emphasize that seaside is not really a standalone HTML model but (as the name says) a canvas which streams markup. >> What we should explore is the possibility of having a HTML model that can be used for streaming, too, instead just the streaming. There could be a lot of synergies here because web frameworks, pillar, PharoJS are all working close to something like that. >> >> my 2 cents, >> >> Norbert

Sept. 17, 2017

Re: [Pharo-users] UFFI problem / doubt

by Esteban Lorenzano

yes, he is using that. but he is porting it yo UFFI (it was NB) and to 64bits and to Chipmunk 7.0 (it was 6.1) Soâ¦ in one of those places, is failing :( Esteban > On 17 Sep 2017, at 13:32, Stephane Ducasse <stepharo.self(a)gmail.com> wrote: > > Look at http://smalltalkhub.com/#!/~estebanlm/Storm > > A small game engine. > > For being able to use this you will need: > > NB capable VM. > Athens. > Chipmunk library. You can download it at Chipmunk > > On Sun, Sep 17, 2017 at 1:31 PM, Stephane Ducasse > <stepharo.self(a)gmail.com> wrote: >> Jean-Baptiste Arnaud extended the Chipmunk binding that esteban did a >> while ago and Damien Pollet did a tutorial >> at ESUG Annecy to show how to use FFI (nativeboost) to call chipmunk >> so it should be around. >> >> Stef >> >> On Sat, Sep 16, 2017 at 2:09 PM, alvaro piorno >> <alvaropiorno100(a)gmail.com> wrote: >>> Hi Stephan, >>> Thanks for your answer, i found the booklet, but it was a draft, so it was >>> incomplete. An example of FFIOpaqueObject would be healpful, i think. >>> Esteban already talked to me, and is helping me with this. >>> What do you mean with "Chipmunk existing bindings" ?. >>> >>> 2017-09-16 4:42 GMT-03:00 Stephane Ducasse <stepharo.self(a)gmail.com>: >>>> >>>> Hi alvaro >>>> >>>> Thanks for the question. If you have any suggestion for the booklet on >>>> UFFI let us know. >>>> I will let esteban anwser because I do not know. >>>> BTW did you check the chipmunk existing bindings? >>>> >>>> Stef >>>> >>>> On Fri, Sep 15, 2017 at 3:42 PM, alvaro piorno >>>> <alvaropiorno100(a)gmail.com> wrote: >>>>> Hi everybody, iÂ´m using UnifiedFFI with Chipmunk2D library . >>>>> I want to get a pointer to a C Structure (cpBody) and set a position to >>>>> it. >>>>> I have a subClass of FFIOpaqueObject to handle pointer. >>>>> >>>>> This is the "new " ffiCall : >>>>> self ffiCall: #(cpBody* cpBodyNew(cpFloat m, cpFloat i)) >>>>> >>>>> And this is a "seter" ffiCall: >>>>> >>>>> self ffiCall: #( void cpBodySetPosition( self, cpVect vector ) ). >>>>> >>>>> I`m not getting any "error" but the position is not updating. >>>>> I tried the same using Chipmunk directly and it works, so there is a >>>>> problem >>>>> with what i have in Pharo. >>>>> Any suggestion? >>>>> Thanks in advance. >>>>> >>>>> Alvaro Piorno >>>> >>> >

Sept. 17, 2017

Re: [Pharo-users] Standalone HTML Model

by Stephane Ducasse

Norbert can you explain the difference between "a HTML model that can be used for streaming, too, instead just the streaming." because indeed in Pillar we have another HTML exporter and I would like to reuse :) On Sun, Sep 17, 2017 at 1:29 PM, Norbert Hartl <norbert(a)hartl.name> wrote: > > >> Am 17.09.2017 um 12:12 schrieb Pierce Ng <pierce(a)samadhiweb.com>: >> >>> On Mon, Sep 11, 2017 at 06:25:55PM +0200, Stephane Ducasse wrote: >>> Hi Pierce Ng >>> How different is the API from Seaside? >>> Because I would like to use it. >>> I like to think modularly :) >> >> Hi Stef, >> >> I modeled it after Seaside's API. However, it was really just a start and needs >> much more work. I created this when investigating writing web apps in Cuis, >> hence the perceived need for a standalone library. Eventually I ran out of time >> and decided to stick to full Seaside loaded into Pharo. >> > So you've built something like seaside canvas? I like to emphasize that seaside is not really a standalone HTML model but (as the name says) a canvas which streams markup. > What we should explore is the possibility of having a HTML model that can be used for streaming, too, instead just the streaming. There could be a lot of synergies here because web frameworks, pillar, PharoJS are all working close to something like that. > > my 2 cents, > > Norbert

Sept. 17, 2017

Re: [Pharo-users] UFFI problem / doubt

by Stephane Ducasse

Look at http://smalltalkhub.com/#!/~estebanlm/Storm A small game engine. For being able to use this you will need: NB capable VM. Athens. Chipmunk library. You can download it at Chipmunk On Sun, Sep 17, 2017 at 1:31 PM, Stephane Ducasse <stepharo.self(a)gmail.com> wrote: > Jean-Baptiste Arnaud extended the Chipmunk binding that esteban did a > while ago and Damien Pollet did a tutorial > at ESUG Annecy to show how to use FFI (nativeboost) to call chipmunk > so it should be around. > > Stef > > On Sat, Sep 16, 2017 at 2:09 PM, alvaro piorno > <alvaropiorno100(a)gmail.com> wrote: >> Hi Stephan, >> Thanks for your answer, i found the booklet, but it was a draft, so it was >> incomplete. An example of FFIOpaqueObject would be healpful, i think. >> Esteban already talked to me, and is helping me with this. >> What do you mean with "Chipmunk existing bindings" ?. >> >> 2017-09-16 4:42 GMT-03:00 Stephane Ducasse <stepharo.self(a)gmail.com>: >>> >>> Hi alvaro >>> >>> Thanks for the question. If you have any suggestion for the booklet on >>> UFFI let us know. >>> I will let esteban anwser because I do not know. >>> BTW did you check the chipmunk existing bindings? >>> >>> Stef >>> >>> On Fri, Sep 15, 2017 at 3:42 PM, alvaro piorno >>> <alvaropiorno100(a)gmail.com> wrote: >>> > Hi everybody, iÂ´m using UnifiedFFI with Chipmunk2D library . >>> > I want to get a pointer to a C Structure (cpBody) and set a position to >>> > it. >>> > I have a subClass of FFIOpaqueObject to handle pointer. >>> > >>> > This is the "new " ffiCall : >>> > self ffiCall: #(cpBody* cpBodyNew(cpFloat m, cpFloat i)) >>> > >>> > And this is a "seter" ffiCall: >>> > >>> > self ffiCall: #( void cpBodySetPosition( self, cpVect vector ) ). >>> > >>> > I`m not getting any "error" but the position is not updating. >>> > I tried the same using Chipmunk directly and it works, so there is a >>> > problem >>> > with what i have in Pharo. >>> > Any suggestion? >>> > Thanks in advance. >>> > >>> > Alvaro Piorno >>> >>

Sept. 17, 2017

Re: [Pharo-users] UFFI problem / doubt

by Stephane Ducasse

Jean-Baptiste Arnaud extended the Chipmunk binding that esteban did a while ago and Damien Pollet did a tutorial at ESUG Annecy to show how to use FFI (nativeboost) to call chipmunk so it should be around. Stef On Sat, Sep 16, 2017 at 2:09 PM, alvaro piorno <alvaropiorno100(a)gmail.com> wrote: > Hi Stephan, > Thanks for your answer, i found the booklet, but it was a draft, so it was > incomplete. An example of FFIOpaqueObject would be healpful, i think. > Esteban already talked to me, and is helping me with this. > What do you mean with "Chipmunk existing bindings" ?. > > 2017-09-16 4:42 GMT-03:00 Stephane Ducasse <stepharo.self(a)gmail.com>: >> >> Hi alvaro >> >> Thanks for the question. If you have any suggestion for the booklet on >> UFFI let us know. >> I will let esteban anwser because I do not know. >> BTW did you check the chipmunk existing bindings? >> >> Stef >> >> On Fri, Sep 15, 2017 at 3:42 PM, alvaro piorno >> <alvaropiorno100(a)gmail.com> wrote: >> > Hi everybody, iÂ´m using UnifiedFFI with Chipmunk2D library . >> > I want to get a pointer to a C Structure (cpBody) and set a position to >> > it. >> > I have a subClass of FFIOpaqueObject to handle pointer. >> > >> > This is the "new " ffiCall : >> > self ffiCall: #(cpBody* cpBodyNew(cpFloat m, cpFloat i)) >> > >> > And this is a "seter" ffiCall: >> > >> > self ffiCall: #( void cpBodySetPosition( self, cpVect vector ) ). >> > >> > I`m not getting any "error" but the position is not updating. >> > I tried the same using Chipmunk directly and it works, so there is a >> > problem >> > with what i have in Pharo. >> > Any suggestion? >> > Thanks in advance. >> > >> > Alvaro Piorno >> >

Sept. 17, 2017

Re: [Pharo-users] Standalone HTML Model

by Norbert Hartl

> Am 17.09.2017 um 12:12 schrieb Pierce Ng <pierce(a)samadhiweb.com>: > >> On Mon, Sep 11, 2017 at 06:25:55PM +0200, Stephane Ducasse wrote: >> Hi Pierce Ng >> How different is the API from Seaside? >> Because I would like to use it. >> I like to think modularly :) > > Hi Stef, > > I modeled it after Seaside's API. However, it was really just a start and needs > much more work. I created this when investigating writing web apps in Cuis, > hence the perceived need for a standalone library. Eventually I ran out of time > and decided to stick to full Seaside loaded into Pharo. > So you've built something like seaside canvas? I like to emphasize that seaside is not really a standalone HTML model but (as the name says) a canvas which streams markup. What we should explore is the possibility of having a HTML model that can be used for streaming, too, instead just the streaming. There could be a lot of synergies here because web frameworks, pillar, PharoJS are all working close to something like that. my 2 cents, Norbert

Sept. 17, 2017

Re: [Pharo-users] Pharo 7 license question

by stephan

On 17-09-17 06:59, Jimmie Houchin wrote: > And the GPL not be viral in my app provided I only use the GPL library > and am not modifying it in my app. > > Do I understand this wrong? Yes. With GPL everything is now GPL. With LGPL, as long as you only link to it, the viral aspect is limited to the library. In Pharo, that means you can use UFFI to connect to LGPL libraries, and you can probably create plugins. Loading smalltalk libraries that are LGPL is not exactly the same as linking, there is no clear boundary between compile-time and run-time, as everything is in the image. That makes the LGPL difficult to interpret in the smalltalk case, and potentially viral. Stephan

Sept. 17, 2017

Re: [Pharo-users] Pharo 7 license question

by stephan

On 17-09-17 12:09, Hilaire wrote: > For library, alternative is LGPL and I read this interesting note: > > One should note that subclassing a Java (or other OO) class licensed > under the LGPL is regarded as a use of an interface of a library > comparable to a function call of a library. It is not regarded as a > modification of the original class. Therefore the subclass does not > fall under the requirements of the LGPL. > > So using a LGPL library, even extending it, does not force the user to > be in the GPL family license. Loading an LGPL library in a smalltalk image is not linking, in the strict interpretation. So no. FSF is aware of this problem Stephan

Sept. 17, 2017