July 2018 - Pharo-users - lists.pharo.org

Re: [Pharo-users] transactions on pharo objects

by tesonep＠gmail.com

Hello, A lot of time ago, we have implemented a basic support for STM in Java (it is written in Scala... dark dark dark ages). We have used aspect oriented programming to intercept the reading and writing of the instance variables of the classes that support the transactions. We kept a dictionary with the values for the transactions in course and a version, and the original object has the "commited" values. On a commit you can apply the changes to the object iff there is no conflict version. I think a similar implementation could be done in Pharo using Slots in a couple of hours. Also, you should use process specific values to detect and handle in witch transaction you are. Depending how much you want to guarantee ACID properties, you will have simpler or complex implementation. In those days, we only wanted only isolation. Sadly, the details of the implementation are in a Master Thesis in Spanish, I doubt they will be useful, but still I can point you to the source code of the solution. By the way, this is still in use as a part of a tool to teach UI design. https://github.com/uqbar-project/arena/tree/master/arena-pot Cheers, Pablo On Tue, Jul 31, 2018 at 8:48 AM Norbert Hartl <norbert(a)hartl.name> wrote: > > > Am 31.07.2018 um 06:57 schrieb Richard O'Keefe <raoknz(a)gmail.com>: > > Basically, what you are talking about is Software Transactional Memory. > According to > https://en.wikipedia.org/wiki/Software_transactional_memory#Smalltalk > there *is* STM support for Pharo at > http://source.lukas-renggli.ch/transactional/ > although the last version there is from 2012, and there have been major > changes to Pharo > since then, so it probably doesn't work any longer. > > You could probably make a TransactionalObject class with a > 'lastTransaction' > instance variable, and a noteChange method that checks if lastTransaction > == > Transaction current, and if not, pushes self -> self shallowCopy onto a > stack > inside Transaction and sets lastTransaction to Transaction current. Then > to > roll back a transaction, peel back original -> backup records from the > stack > and do original copyFrom: backup for each of them. > > Please don't ask me to think about combining this with concurrency. > > You could delegate the transaction list and other objects to a process > specific variable. But the biggest problem with a copy approach is that all > identity checks fail with the copied objects > > Norbert > > On 31 July 2018 at 01:16, Peter UhnÃ¡k <i.uhnak(a)gmail.com> wrote: > >> Hi, >> >> is there some library or approach how to do transactions in pharo? >> And I don't mean database transactions, but directly in memory on Pharo >> objects... e.g. >> >> p := Person new. >> >> transaction do: [ >> p name: 'Nobody'. >> p age: 70. >> ] on: Error do: [ >> transaction rollback. >> ]. >> >> self assert: p name equals: 'Nobody'. >> self assert: p age equals: 70. >> >> transaction do: [ >> p name: 'Somebody'. >> p age: 1 / 0. >> ] on: Error do: [ >> transaction rollback. >> ]. >> >> self assert: p name equals: 'Nobody'. >> self assert: p age equals: 70. >> >> Any pointers appreciated. >> >> Thanks, >> Peter >> > > -- Pablo Tesone. tesonep(a)gmail.com

July 31, 2018

Re: [Pharo-users] a basic hash question

by John Pfersich

This inconsistent naming convention has me wishing for an updated ANSI Standard. There has been much change, wasnât that standard from the 90s? /ââââââââââââââââââââ/ For encrypted mail use jgpfersich(a)protonmail.com Get a free account at ProtonMail.com Web: www.objectnets.net and www.objectnets.org > On Jul 30, 2018, at 22:59, Richard O'Keefe <raoknz(a)gmail.com> wrote: > > I do not think that (1 to: 4) and #(1 2 3 4) should be equal. > Let me put it a little more strongly: it's a bug. > Taking > a := 1 to: 4. > b := Array withAll: a. > c := OrderedCollection withAll: b. > in the two other Smalltalk systems I just tried, > no two of these are equal. This is what the ANSI > Smalltalk standard requires. Ceteris paribus, > two sequences are equivalent if and only if > 1. they are instance of the same class > 2. they have the same size > 3. corresponding elements are equivalent. > It is fairly common for Smalltalk systems to distinguish > between "these sequences are equivalent" and "these > sequences have the same elements in the same order", > with no consensus on the name of the second method. > One calls it #sameContentsAs:, Squeak #hasEqualElements:. > > >> On 31 July 2018 at 00:07, werner kassens <wkassens(a)libello.com> wrote: >> Hi, >> i guess i can subsume almost everything i know about hashes in one sentence: >> it is my understanding that two objects that are equal (obj1=obj2. >> -->true) have to have the same hash value (which is used for some >> collection types), whereas objects where obj1=obj2 returns false should >> have different hash values although it may happen that they have the >> same one. >> >> now here things don't turn out exactly like that: >> (1 to:4) = #(1 2 3 4). "true" >> (1 to:4)hash = #(1 2 3 4)hash. "false" >> well ok, actually these results make - pfffh, in a certain way - sense >> to me, but i wonder what arguments the people in the know would use to >> defend that result, if i would have another opinion? >> werner >

July 31, 2018

Re: [Pharo-users] transactions on pharo objects

by Norbert Hartl

> Am 31.07.2018 um 06:57 schrieb Richard O'Keefe <raoknz(a)gmail.com>: > > Basically, what you are talking about is Software Transactional Memory. > According to https://en.wikipedia.org/wiki/Software_transactional_memory#Smalltalk > there *is* STM support for Pharo at > http://source.lukas-renggli.ch/transactional/ > although the last version there is from 2012, and there have been major changes to Pharo > since then, so it probably doesn't work any longer. > > You could probably make a TransactionalObject class with a 'lastTransaction' > instance variable, and a noteChange method that checks if lastTransaction == > Transaction current, and if not, pushes self -> self shallowCopy onto a stack > inside Transaction and sets lastTransaction to Transaction current. Then to > roll back a transaction, peel back original -> backup records from the stack > and do original copyFrom: backup for each of them. > > Please don't ask me to think about combining this with concurrency. > You could delegate the transaction list and other objects to a process specific variable. But the biggest problem with a copy approach is that all identity checks fail with the copied objects Norbert >> On 31 July 2018 at 01:16, Peter UhnÃ¡k <i.uhnak(a)gmail.com> wrote: >> Hi, >> >> is there some library or approach how to do transactions in pharo? >> And I don't mean database transactions, but directly in memory on Pharo objects... e.g. >> >> p := Person new. >> >> transaction do: [ >> p name: 'Nobody'. >> p age: 70. >> ] on: Error do: [ >> transaction rollback. >> ]. >> >> self assert: p name equals: 'Nobody'. >> self assert: p age equals: 70. >> >> transaction do: [ >> p name: 'Somebody'. >> p age: 1 / 0. >> ] on: Error do: [ >> transaction rollback. >> ]. >> >> self assert: p name equals: 'Nobody'. >> self assert: p age equals: 70. >> >> Any pointers appreciated. >> >> Thanks, >> Peter >

July 31, 2018

Re: [Pharo-users] transactions on pharo objects

by Norbert Hartl

> Am 30.07.2018 um 16:02 schrieb Sean P. DeNigris <sean(a)clipperadams.com>: > > Peter UhnÃ¡k wrote >> is there some library or approach how to do transactionsâ¦ directly in >> memory on Pharo >> objects > > Magritte? It uses the Memento pattern to verify all changes before > committing to real object. > This only works for simple use cases where there is one flow that creates the memento and keeps it like in a web form. The problem with the memento is that it is hard to make all objects give a memento when they are aquired. Norbert > > ----- > Cheers, > Sean > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html >

July 31, 2018

Re: [Pharo-users] transactions on pharo objects

by Norbert Hartl

> Am 30.07.2018 um 15:16 schrieb Peter UhnÃ¡k <i.uhnak(a)gmail.com>: > > Hi, > > is there some library or approach how to do transactions in pharo? > And I don't mean database transactions, but directly in memory on Pharo objects... e.g. > > p := Person new. > > transaction do: [ > p name: 'Nobody'. > p age: 70. > ] on: Error do: [ > transaction rollback. > ]. > > self assert: p name equals: 'Nobody'. > self assert: p age equals: 70. > > transaction do: [ > p name: 'Somebody'. > p age: 1 / 0. > ] on: Error do: [ > transaction rollback. > ]. > > self assert: p name equals: 'Nobody'. > self assert: p age equals: 70. > > Any pointers appreciated. > Should this work image wide or per process? Norbert

July 31, 2018

Re: [Pharo-users] a basic hash question

by Richard O'Keefe

+1 to what Andreas Valloud (Mr "how to hash in Smalltalk") said. On 31 July 2018 at 10:34, Andres Valloud <avalloud(a)smalltalk.comcastbiz.net> wrote: > The interval > > 1 to: (10 raisedTo: 100) > > can be created just fine, yet hashing its elements won't compute. > > A generous interpretation of the intent of #=, where any wisp of > equivalence is promoted to full fledged equality, is problematic in the > long run. Here's another one: > > 17/20 = 0.85, therefore (17/20) hash = 0.85 hash > > Never mind there's no floating point value that is equal to 17/20 in the > first place. It just snowballs from there. For instance, should > collections like these be equal? > > (0.0 to: 1.0 by: 0.3) = #(0.0 0.3 0.6 0.9) > > Sometimes it's just better if different things stay different. > > Andres. > > > On 7/30/18 8:27 , Henrik Sperre Johansen wrote: > >> jgfoster wrote >> >>> Werner, >>> >>> I would say that you are right, this is a problem. A (not un-common) >>> source of subtle bugs in Smalltalk is missing this rule that equivalent >>> objects must have the same hash. In GemStone the objects are not >>> equivalent (Iâm not arguing that this is right, just that it avoids the >>> problem you identify). >>> >>> I wonder what would happen if the hash comparison were added to the >>> equivalence operator (#â=â)! >>> >>> James >>> >>> On Jul 30, 2018, at 5:07 AM, werner kassens < >>>> >>> >> wkassens@ >>> >> >> > wrote: >>> >>>> >>>> Hi, >>>> i guess i can subsume almost everything i know about hashes in one >>>> sentence: >>>> it is my understanding that two objects that are equal (obj1=obj2. >>>> -->true) have to have the same hash value (which is used for some >>>> collection types), whereas objects where obj1=obj2 returns false should >>>> have different hash values although it may happen that they have the >>>> same one. >>>> >>>> now here things don't turn out exactly like that: >>>> (1 to:4) = #(1 2 3 4). "true" >>>> (1 to:4)hash = #(1 2 3 4)hash. "false" >>>> well ok, actually these results make - pfffh, in a certain way - sense >>>> to me, but i wonder what arguments the people in the know would use to >>>> defend that result, if i would have another opinion? >>>> werner >>>> >>>> >>>> >>>> >> +1, this is a bug. >> Either Interval >> #hash needs to change*, or the equivalency be broken**. >> >> Fun fact: >> #(1 2 3 4) = #[1 2 3 4] false >> You'd think they might be more similar than an Interval and an Array, or a >> LinkedList containing ValueLinks and an Interval, but no ;) >> >> Cheers, >> Henry >> >> *Interval >> hash >> "Hash is reimplemented because = is implemented. >> Since we are equivalent to other collections of our species, we >> must also >> hash equivalently" >> | hash | >> >> hash := self species hash. >> start to: stop by: step do: [:element | hash := (hash + element >> hash) >> hashMultiply]. >> ^hash >> >> **both ways, which is challenging, if one neither wants to: >> - change species of Interval (which has bad consequences) >> - add hash comparison to sequenceablecollection = (which slows it down >> even >> further) >> >> >> >> >> -- >> Sent from: http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html >> >> . >> >> >

July 31, 2018

Re: [Pharo-users] a basic hash question

by Richard O'Keefe

I do not think that (1 to: 4) and #(1 2 3 4) should be equal. Let me put it a little more strongly: it's a bug. Taking a := 1 to: 4. b := Array withAll: a. c := OrderedCollection withAll: b. in the two other Smalltalk systems I just tried, no two of these are equal. This is what the ANSI Smalltalk standard requires. Ceteris paribus, two sequences are equivalent if and only if 1. they are instance of the same class 2. they have the same size 3. corresponding elements are equivalent. It is fairly common for Smalltalk systems to distinguish between "these sequences are equivalent" and "these sequences have the same elements in the same order", with no consensus on the name of the second method. One calls it #sameContentsAs:, Squeak #hasEqualElements:. On 31 July 2018 at 00:07, werner kassens <wkassens(a)libello.com> wrote: > Hi, > i guess i can subsume almost everything i know about hashes in one > sentence: > it is my understanding that two objects that are equal (obj1=obj2. > -->true) have to have the same hash value (which is used for some > collection types), whereas objects where obj1=obj2 returns false should > have different hash values although it may happen that they have the > same one. > > now here things don't turn out exactly like that: > (1 to:4) = #(1 2 3 4). "true" > (1 to:4)hash = #(1 2 3 4)hash. "false" > well ok, actually these results make - pfffh, in a certain way - sense > to me, but i wonder what arguments the people in the know would use to > defend that result, if i would have another opinion? > werner > > >

July 31, 2018

Re: [Pharo-users] a basic hash question

by werner kassens

Hi Andres, <g> that is the kind of argument i was looking for, as i thought i would have a similar situation as in my small example and wondered whether i could keep my slightly incongruent definition of #= and #hash, but in a way your examples show that my problem is different, iow i should change my implementation & i have changed it. werner On 07/31/2018 12:34 AM, Andres Valloud wrote: > The interval > > Â Â Â Â 1 to: (10 raisedTo: 100) > > can be created just fine, yet hashing its elements won't compute. > > A generous interpretation of the intent of #=, where any wisp of > equivalence is promoted to full fledged equality, is problematic in > the long run.Â Here's another one: > > Â Â Â Â 17/20 = 0.85, therefore (17/20) hash = 0.85 hash > > Never mind there's no floating point value that is equal to 17/20 in > the first place.Â It just snowballs from there.Â For instance, should > collections like these be equal? > > Â Â Â Â (0.0 to: 1.0 by: 0.3) = #(0.0 0.3 0.6 0.9) > > Sometimes it's just better if different things stay different. > > Andres.

July 31, 2018

Re: [Pharo-users] transactions on pharo objects

by Richard O'Keefe

Basically, what you are talking about is Software Transactional Memory. According to https://en.wikipedia.org/wiki/Software_transactional_memory#Smalltalk there *is* STM support for Pharo at http://source.lukas-renggli.ch/transactional/ although the last version there is from 2012, and there have been major changes to Pharo since then, so it probably doesn't work any longer. You could probably make a TransactionalObject class with a 'lastTransaction' instance variable, and a noteChange method that checks if lastTransaction == Transaction current, and if not, pushes self -> self shallowCopy onto a stack inside Transaction and sets lastTransaction to Transaction current. Then to roll back a transaction, peel back original -> backup records from the stack and do original copyFrom: backup for each of them. Please don't ask me to think about combining this with concurrency. On 31 July 2018 at 01:16, Peter UhnÃ¡k <i.uhnak(a)gmail.com> wrote: > Hi, > > is there some library or approach how to do transactions in pharo? > And I don't mean database transactions, but directly in memory on Pharo > objects... e.g. > > p := Person new. > > transaction do: [ > p name: 'Nobody'. > p age: 70. > ] on: Error do: [ > transaction rollback. > ]. > > self assert: p name equals: 'Nobody'. > self assert: p age equals: 70. > > transaction do: [ > p name: 'Somebody'. > p age: 1 / 0. > ] on: Error do: [ > transaction rollback. > ]. > > self assert: p name equals: 'Nobody'. > self assert: p age equals: 70. > > Any pointers appreciated. > > Thanks, > Peter >

July 31, 2018

Re: [Pharo-users] a basic hash question

by Andres Valloud

The interval 1 to: (10 raisedTo: 100) can be created just fine, yet hashing its elements won't compute. A generous interpretation of the intent of #=, where any wisp of equivalence is promoted to full fledged equality, is problematic in the long run. Here's another one: 17/20 = 0.85, therefore (17/20) hash = 0.85 hash Never mind there's no floating point value that is equal to 17/20 in the first place. It just snowballs from there. For instance, should collections like these be equal? (0.0 to: 1.0 by: 0.3) = #(0.0 0.3 0.6 0.9) Sometimes it's just better if different things stay different. Andres. On 7/30/18 8:27 , Henrik Sperre Johansen wrote: > jgfoster wrote >> Werner, >> >> I would say that you are right, this is a problem. A (not un-common) >> source of subtle bugs in Smalltalk is missing this rule that equivalent >> objects must have the same hash. In GemStone the objects are not >> equivalent (Iâm not arguing that this is right, just that it avoids the >> problem you identify). >> >> I wonder what would happen if the hash comparison were added to the >> equivalence operator (#â=â)! >> >> James >> >>> On Jul 30, 2018, at 5:07 AM, werner kassens < > >> wkassens@ > >> > wrote: >>> >>> Hi, >>> i guess i can subsume almost everything i know about hashes in one >>> sentence: >>> it is my understanding that two objects that are equal (obj1=obj2. >>> -->true) have to have the same hash value (which is used for some >>> collection types), whereas objects where obj1=obj2 returns false should >>> have different hash values although it may happen that they have the >>> same one. >>> >>> now here things don't turn out exactly like that: >>> (1 to:4) = #(1 2 3 4). "true" >>> (1 to:4)hash = #(1 2 3 4)hash. "false" >>> well ok, actually these results make - pfffh, in a certain way - sense >>> to me, but i wonder what arguments the people in the know would use to >>> defend that result, if i would have another opinion? >>> werner >>> >>> >>> > > +1, this is a bug. > Either Interval >> #hash needs to change*, or the equivalency be broken**. > > Fun fact: > #(1 2 3 4) = #[1 2 3 4] false > You'd think they might be more similar than an Interval and an Array, or a > LinkedList containing ValueLinks and an Interval, but no ;) > > Cheers, > Henry > > *Interval >> hash > "Hash is reimplemented because = is implemented. > Since we are equivalent to other collections of our species, we must also > hash equivalently" > | hash | > > hash := self species hash. > start to: stop by: step do: [:element | hash := (hash + element hash) > hashMultiply]. > ^hash > > **both ways, which is challenging, if one neither wants to: > - change species of Interval (which has bad consequences) > - add hash comparison to sequenceablecollection = (which slows it down even > further) > > > > > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html > > . >

July 30, 2018