- Pharo-dev - lists.pharo.org

Re: [Pharo-dev] WeakMessageSend is broken?

by Aliaksei Syrel

Submitted an issue https://pharo.fogbugz.com/f/cases/17906/WeakMessageSend-is-broken Cheers, Alex On Mon, Mar 28, 2016 at 9:29 PM, stepharo <stepharo(a)free.fr> wrote: > Hi aliaksei > > I do not know. Now I know that last week guille was trying to understand > memory leaks > and they all were around Weak structures > > Stef > > Le 28/3/16 21:08, Aliaksei Syrel a Ã©crit : > > Assume we have the following weak message send: > > message := WeakMessageSend >> receiver: Morph new >> selector: #color. >> 100000 timesRepeat: [ message value ] > > > If you execute it in latest Pharo and VM you will (almost 100% sure) get > DNU. (but it should just return nil if receiver is garbage collected) > > How can I ensure that it selector is not being sent to nil? > > <http://ws.stfx.eu/6QOWBLQENVIY>http://ws.stfx.eu/6QOWBLQENVIY > > Am I doing something wrong? > > Cheers, > Alex > > >

March 28, 2016

Re: [Pharo-dev] WeakMessageSend is broken?

by stepharo

Hi aliaksei I do not know. Now I know that last week guille was trying to understand memory leaks and they all were around Weak structures Stef Le 28/3/16 21:08, Aliaksei Syrel a Ã©crit : > Assume we have the following weak message send: > > message := WeakMessageSend > receiver: Morph new > selector: #color. > 100000 timesRepeat: [ message value ] > > > If you execute it in latest Pharo and VM you will (almost 100% sure) > get DNU. (but it should just return nil if receiver is garbage collected) > > How can I ensure that it selector is not being sent to nil? > > http://ws.stfx.eu/6QOWBLQENVIY > > Am I doing something wrong? > _ > _ > Cheers, > Alex

March 28, 2016

WeakMessageSend is broken?

by Aliaksei Syrel

Assume we have the following weak message send: message := WeakMessageSend > receiver: Morph new > selector: #color. > 100000 timesRepeat: [ message value ] If you execute it in latest Pharo and VM you will (almost 100% sure) get DNU. (but it should just return nil if receiver is garbage collected) How can I ensure that it selector is not being sent to nil? http://ws.stfx.eu/6QOWBLQENVIY Am I doing something wrong? Cheers, Alex

March 28, 2016

Re: [Pharo-dev] problem with slots on proxies

by Tudor Girba

Hi, > On Mar 27, 2016, at 6:47 PM, Denis Kudriashov <dionisiydk(a)gmail.com> wrote: > > Hi > > 2016-03-26 21:54 GMT+01:00 Tudor Girba <tudor(a)tudorgirba.com>: > Hi, > > I am trying to play with some proxies that inherit from ProtoObject have slots. For example, I try to do this, but I get an error: > > ProtoObject subclass: #BBB > instanceVariableNames: 'variable' > classVariableNames: '' > package: 'BBB'. > BBB class allSlots first read: BBB new > > The problem is that the slot sends messages to its parent object, such as: > IndexedSlot>>read: anObject > ^ anObject instVarAt: index > > This is a problem because a similar scenario happens also when you open a debugger on a proxy that tries to communicate with a remote image. > > I think that instead of sending messages to the receiver, we should make the slot work with the mirror primitives. > > For example, we would rewrite the read: method like: > IndexedSlot>>read: anObject > ^ thisContext object: anObject instVarAt: index > > Do you agree? Can I open an issue for this? I think this is important. > > Yes, please. Ok, I created the issue and the associated fix: https://pharo.fogbugz.com/f/cases/17899/Slots-should-not-send-messages-to-t… > But does not thisContext degrades performance? > Guille extracted mirror primitives to separate class MirrorPrimitives (project Mirror in Pharo repo). I think we should start to use it and move away from context. I know about the Mirror project, but as long as it is not part of the Core, we cannot use it for Slots :). Cheers, Doru -- www.tudorgirba.com www.feenk.com "We are all great at making mistakes."

March 27, 2016

Re: [Pharo-dev] [Ann] New version of Ghost

by Denis Kudriashov

2016-03-24 20:27 GMT+01:00 stepharo <stepharo(a)free.fr>: > > - ghost ghostClass "returns actual class of receiver" > > when I read ghostClass I expect getting the class > implementing the proxy not the proxied class. > So I'm confused. Now I am confused too :)) ghost ghostClass "=> anObjectGhost (or subclass)"

March 27, 2016

Re: [Pharo-dev] [Ann] New version of Ghost

by Denis Kudriashov

2016-03-25 1:34 GMT+01:00 Mariano Martinez Peck <marianopeck(a)gmail.com>: > Good example where I use viruses is stubbing real objects in Mocketry. I >> will announce it soon. You can do this: >> >> rect := 0@0 corner: 2@3 >> rect stub area willReturn: 1000. >> >> rect area "=> 1000". >> rect origin "=> 0@0". >> >> >> > OK..I should take a look to the code...but...it looks to me that stub the > only thing it does is to create the proxy and then the #become: ?? > No. Stub should intercept all self-sends. Classic proxy can't do this. Better example to show self-send stubbing: rect := 0@0 corner: 2@3 rect stub width willReturn: 1000. rect area "=> 3000" (area is width and height multiplication).

March 27, 2016

Re: [Pharo-dev] problem with slots on proxies

by Denis Kudriashov

Hi 2016-03-26 21:54 GMT+01:00 Tudor Girba <tudor(a)tudorgirba.com>: > Hi, > > I am trying to play with some proxies that inherit from ProtoObject have > slots. For example, I try to do this, but I get an error: > > ProtoObject subclass: #BBB > instanceVariableNames: 'variable' > classVariableNames: '' > package: 'BBB'. > BBB class allSlots first read: BBB new > > The problem is that the slot sends messages to its parent object, such as: > IndexedSlot>>read: anObject > ^ anObject instVarAt: index > > This is a problem because a similar scenario happens also when you open a > debugger on a proxy that tries to communicate with a remote image. > > I think that instead of sending messages to the receiver, we should make > the slot work with the mirror primitives. > > For example, we would rewrite the read: method like: > IndexedSlot>>read: anObject > ^ thisContext object: anObject instVarAt: index > > Do you agree? Can I open an issue for this? I think this is important. > Yes, please. But does not thisContext degrades performance? Guille extracted mirror primitives to separate class MirrorPrimitives (project Mirror in Pharo repo). I think we should start to use it and move away from context.

March 27, 2016

Re: [Pharo-dev] Class comments need review (English speakers welcome)

by Cyril Ferlicot Delbecque

On 27/03/2016 11:42, stepharo wrote: > Cool cyril > > What I was thinking is to copy what we wrote in the Pharo by example > > For each class I read the PBE part on Collection's chapter to see if I covered enough things in the comment. -- Cyril Ferlicot http://www.synectique.eu 165 Avenue Bretagne Lille 59000 France

March 27, 2016

Re: [Pharo-dev] Class comments need review (English speakers welcome)

by stepharo

Cool cyril What I was thinking is to copy what we wrote in the Pharo by example Le 26/3/16 21:53, Cyril Ferlicot Delbecque a Ã©crit : > Hi! > > I begun to improve the comments of Collection Sequenceable's classes. I > would like the community to review few things: > - The English! Since I learned English on video games let's say that my > level in far from been perfect :) I most probably made mistakes and I > would like English speakers to review that part. > - The content. Maybe I wrote bad/useless informations? Maybe I forgot > something? Maybe there is an implementation detail you know and that > deserve to be in the comment? Maybe you have a cool example to add? > - I did not improved the comment of some classes because I don't have > enough knowledge on them. (Mostly SharedQueue, SparseLargeTable, > SparseLargeArray) It would be welcome that people improve those > comments. Or if you don't want to do it you can send me a message with a > description of the class and example of usages in the system (since I > think that they are used by the core of Pharo). > > I mostly changed those classes comment: > - Interval > - LinkedList > - OrderedCollection > - SortedCollection > - OrderedDictionary > - OrderedIdentityDictionary > > If you want to help you can merge this slice from Pharo50Inbox: > > Name: > SLICE-Issue-16945-Improve-Collection-Sequenceable-comments-CyrilFerlicot.3 > Author: CyrilFerlicot > Time: 26 March 2016, 9:50:43.042998 pm > UUID: 54adf1ee-27e4-49e3-ab07-2552486f5f62 > Ancestors: > SLICE-Issue-16945-Improve-Collection-Sequenceable-comments-CyrilFerlicot.2 > Dependencies: Collections-Sequenceable-CyrilFerlicot.238 > > add example to OrderedIdentityDictionary > > > Thank you in advance for your help! Let's make Pharo a user friendly > environment. :) >

March 27, 2016

problem with slots on proxies

by Tudor Girba

Hi, I am trying to play with some proxies that inherit from ProtoObject have slots. For example, I try to do this, but I get an error: ProtoObject subclass: #BBB instanceVariableNames: 'variable' classVariableNames: '' package: 'BBB'. BBB class allSlots first read: BBB new The problem is that the slot sends messages to its parent object, such as: IndexedSlot>>read: anObject ^ anObject instVarAt: index This is a problem because a similar scenario happens also when you open a debugger on a proxy that tries to communicate with a remote image. I think that instead of sending messages to the receiver, we should make the slot work with the mirror primitives. For example, we would rewrite the read: method like: IndexedSlot>>read: anObject ^ thisContext object: anObject instVarAt: index Do you agree? Can I open an issue for this? I think this is important. Cheers, Doru -- www.tudorgirba.com www.feenk.com "If you can't say why something is relevant, it probably isn't."

March 26, 2016