January 2013 - Pharo-users - lists.pharo.org

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Jesus NuÃ±ez

Yes Ok, I got it. The question is that the #value: selector is still there, with the old code. Sorry, I should have check the new code directly in the repository and compare it. Thank you. 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> > On 12 January 2013 19:00, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: > > BTW I also tried returning with ";yourself" but no luck in the test I > made. > > > > > i dont know what you trying to change there. this code works fine, because > if it would not, then nothing else would work at all. > > > > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> > >> > >> Ok, but happens that #value: uses #fieldAt: was inherited from > >> NBExternalStructure, which is no longer in the hierarchy. But I > understand > >> that it will no longer work with the change I made, any idea? > >> > >> > >> Best Regards, > >> > >> Jesus > >> > >> > >> 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> > >>> > >>> No, it should be > >>> NBExternalAddress value: addr > >>> because method should answer an instance of NBExternalAddress > >>> not integer. > >>> > >>> On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> > wrote: > >>> > Sorry. NBExternalAddress new ulongAt: adress > >>> > > >>> > > >>> > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> > >>> >> > >>> >> I have made a small change in the code because NBExternalAdress does > >>> >> not > >>> >> inherit from NBExternalStructure anymore. It is safe?. I mean so far > >>> >> no > >>> >> problems but my understanding is limited so I haven't make any > commit. > >>> >> > >>> >> getGlobalSymbolPointer: symbolName > >>> >> | bytes addr | > >>> >> bytes := ByteArray new: 8. > >>> >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. > >>> >> addr := bytes unsignedLongAt: 1 bigEndian: false. > >>> >> addr = 0 > >>> >> ifTrue: [self error: 'failed to get a symbol address: ' , > symbolName]. > >>> >> +^ NBExternalAddress ulongAt: addr > >>> >> -^ NBExternalAddress value: addr > >>> >> > >>> >> Cheers > >>> > > >>> > > >>> > >>> > >>> > >>> -- > >>> Best regards, > >>> Igor Stasenko. > >>> > >> > > > > > > -- > Best regards, > Igor Stasenko. > >

Jan. 12, 2013

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Igor Stasenko

On 12 January 2013 19:00, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: > BTW I also tried returning with ";yourself" but no luck in the test I made. > > i dont know what you trying to change there. this code works fine, because if it would not, then nothing else would work at all. > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> >> >> Ok, but happens that #value: uses #fieldAt: was inherited from >> NBExternalStructure, which is no longer in the hierarchy. But I understand >> that it will no longer work with the change I made, any idea? >> >> >> Best Regards, >> >> Jesus >> >> >> 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> >>> >>> No, it should be >>> NBExternalAddress value: addr >>> because method should answer an instance of NBExternalAddress >>> not integer. >>> >>> On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: >>> > Sorry. NBExternalAddress new ulongAt: adress >>> > >>> > >>> > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> >>> >> >>> >> I have made a small change in the code because NBExternalAdress does >>> >> not >>> >> inherit from NBExternalStructure anymore. It is safe?. I mean so far >>> >> no >>> >> problems but my understanding is limited so I haven't make any commit. >>> >> >>> >> getGlobalSymbolPointer: symbolName >>> >> | bytes addr | >>> >> bytes := ByteArray new: 8. >>> >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. >>> >> addr := bytes unsignedLongAt: 1 bigEndian: false. >>> >> addr = 0 >>> >> ifTrue: [self error: 'failed to get a symbol address: ' , symbolName]. >>> >> +^ NBExternalAddress ulongAt: addr >>> >> -^ NBExternalAddress value: addr >>> >> >>> >> Cheers >>> > >>> > >>> >>> >>> >>> -- >>> Best regards, >>> Igor Stasenko. >>> >> > -- Best regards, Igor Stasenko.

Jan. 12, 2013

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Igor Stasenko

On 12 January 2013 18:57, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: > Ok, but happens that #value: uses #fieldAt: was inherited from > NBExternalStructure, which is no longer in the hierarchy. But I understand > that it will no longer work with the change I made, any idea? > Yes, because address is not a structure. Initially it was like that to reuse some code pieces, but it not worth it. > > Best Regards, > > Jesus > > > 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> >> >> No, it should be >> NBExternalAddress value: addr >> because method should answer an instance of NBExternalAddress >> not integer. >> >> On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: >> > Sorry. NBExternalAddress new ulongAt: adress >> > >> > >> > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> >> >> >> >> I have made a small change in the code because NBExternalAdress does >> >> not >> >> inherit from NBExternalStructure anymore. It is safe?. I mean so far no >> >> problems but my understanding is limited so I haven't make any commit. >> >> >> >> getGlobalSymbolPointer: symbolName >> >> | bytes addr | >> >> bytes := ByteArray new: 8. >> >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. >> >> addr := bytes unsignedLongAt: 1 bigEndian: false. >> >> addr = 0 >> >> ifTrue: [self error: 'failed to get a symbol address: ' , symbolName]. >> >> +^ NBExternalAddress ulongAt: addr >> >> -^ NBExternalAddress value: addr >> >> >> >> Cheers >> > >> > >> >> >> >> -- >> Best regards, >> Igor Stasenko. >> > -- Best regards, Igor Stasenko.

Jan. 12, 2013

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Jesus NuÃ±ez

BTW I also tried returning with ";yourself" but no luck in the test I made. 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> > Ok, but happens that #value: uses #fieldAt: was inherited from > NBExternalStructure, which is no longer in the hierarchy. But I understand > that it will no longer work with the change I made, any idea? > > > Best Regards, > > Jesus > > > 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> > >> No, it should be >> NBExternalAddress value: addr >> because method should answer an instance of NBExternalAddress >> not integer. >> >> On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: >> > Sorry. NBExternalAddress new ulongAt: adress >> > >> > >> > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> >> >> >> >> I have made a small change in the code because NBExternalAdress does >> not >> >> inherit from NBExternalStructure anymore. It is safe?. I mean so far no >> >> problems but my understanding is limited so I haven't make any commit. >> >> >> >> getGlobalSymbolPointer: symbolName >> >> | bytes addr | >> >> bytes := ByteArray new: 8. >> >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. >> >> addr := bytes unsignedLongAt: 1 bigEndian: false. >> >> addr = 0 >> >> ifTrue: [self error: 'failed to get a symbol address: ' , symbolName]. >> >> +^ NBExternalAddress ulongAt: addr >> >> -^ NBExternalAddress value: addr >> >> >> >> Cheers >> > >> > >> >> >> >> -- >> Best regards, >> Igor Stasenko. >> >> >

Jan. 12, 2013

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Jesus NuÃ±ez

Ok, but happens that #value: uses #fieldAt: was inherited from NBExternalStructure, which is no longer in the hierarchy. But I understand that it will no longer work with the change I made, any idea? Best Regards, Jesus 2013/1/12 Igor Stasenko <siguctua(a)gmail.com> > No, it should be > NBExternalAddress value: addr > because method should answer an instance of NBExternalAddress > not integer. > > On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: > > Sorry. NBExternalAddress new ulongAt: adress > > > > > > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> > >> > >> I have made a small change in the code because NBExternalAdress does not > >> inherit from NBExternalStructure anymore. It is safe?. I mean so far no > >> problems but my understanding is limited so I haven't make any commit. > >> > >> getGlobalSymbolPointer: symbolName > >> | bytes addr | > >> bytes := ByteArray new: 8. > >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. > >> addr := bytes unsignedLongAt: 1 bigEndian: false. > >> addr = 0 > >> ifTrue: [self error: 'failed to get a symbol address: ' , symbolName]. > >> +^ NBExternalAddress ulongAt: addr > >> -^ NBExternalAddress value: addr > >> > >> Cheers > > > > > > > > -- > Best regards, > Igor Stasenko. > >

Jan. 12, 2013

Re: [Pharo-users] Change in NativeLinux32>>getGlobalSymbolPointer

by Igor Stasenko

No, it should be NBExternalAddress value: addr because method should answer an instance of NBExternalAddress not integer. On 12 January 2013 15:25, Jesus NuÃ±ez <poissonbreaker(a)gmail.com> wrote: > Sorry. NBExternalAddress new ulongAt: adress > > > 2013/1/12 Jesus NuÃ±ez <poissonbreaker(a)gmail.com> >> >> I have made a small change in the code because NBExternalAdress does not >> inherit from NBExternalStructure anymore. It is safe?. I mean so far no >> problems but my understanding is limited so I haven't make any commit. >> >> getGlobalSymbolPointer: symbolName >> | bytes addr | >> bytes := ByteArray new: 8. >> self bootstrapGetGlobalSymbolPointer: symbolName into: bytes. >> addr := bytes unsignedLongAt: 1 bigEndian: false. >> addr = 0 >> ifTrue: [self error: 'failed to get a symbol address: ' , symbolName]. >> +^ NBExternalAddress ulongAt: addr >> -^ NBExternalAddress value: addr >> >> Cheers > > -- Best regards, Igor Stasenko.

Jan. 12, 2013

Re: [Pharo-users] Using Utility classes - follow up to How do I save contents of Workspace into a collection?

by Ben Coman

Ben Coman wrote: > Frank Church wrote: >> Igor made this comment in relation to this thread. >> >> >>> indeed. >>> I wonder why you (Frank) need such complicated way for doing things. >>> If you need a reusable piece of code, which you want to use more >>> than 1 time, >>> classes is the best placeholder for them.. >>> >> >> >>> Even if you put things at class side, it helps a lot.. so then you can >>> just type: >>> >> >> >>> MyUtilsClass doThis >>> >> >> >>> MyUtilsClass doThat >>> >> >> When I was creating the script to automatically add instance variables >> and accessors Stack Overflow 14234003, Mef mentioned that if rather >> than run the method against an instance of an object, I could attach >> the method to a class, namely ClassDescription. >> >> >>> You could add it to ClassDescription itself, then you wouldn't even >>> have to pass the class as parameter (because self is the class >>> then). You could call ClassX addVariablesNamed: aCollection in this >>> case >>> >> >> Is ClassX an example of the MyUtils class Igor mentions? I tried that >> option using ClassDescription but got an MNU error.. >> >> e.g. right now I use the first option in this manner: >> >> | anyVariable | >> anyVariable := ClassType new. >> anyVariable addVarsAndTheirAccessors: #('aaaa' 'bbbb' 'cccc') >> >> How would I attach the method to a utilities class and >> ClassDescription for that matter as it would then apply to all >> classes? >> >> > I don't quite understand you question, but perhaps executing the > following would shed some light... > Morph browseHierarchy > Morph class browseHierarchy > > Define the following method... > ClassDescription>>mytest > self confirm: 'nothing'. > > Then execute... > Morph mytest > > cheers -ben > > btw, do you know that the class definition that you typically see in the browser is just a message being sent to the superclass ? For example, in a Workspace or any method you can execute the following.... ------ ivars := 'aaa bbb ccc'. newclass := #MyTest. Object subclass: newclass instanceVariableNames: ivars classVariableNames: '' poolDictionaries: '' category: 'my-playground' ------

Jan. 12, 2013

Re: [Pharo-users] Is there a way to automatically format a method during the compilation phase or immediately after?

by Paul DeBruicker

Which version of Pharo? in 1.1-1.4 open the world menu, click system, click settings, type format, click the top 5 checkboxes. Probably the same in Pharo 2 too. On 01/12/2013 08:29 AM, Ben Coman wrote: > Frank Church wrote: >> I followed the advice from this question on stackoverflow 14234003 and >> from there created this simple method to add instance variables and >> the basic accessors in one go. >> >> addVarsAndTheirAccessors: aCollection on: aClass >> aCollection do: [ :each | aClass addInstVarNamed: each; >> compile: each, ' ^ ', each classified: 'accessing'; >> compile: each,': aByteString ',each, ' := aByteString' >> classified:'accessing' ] >> >> Currently I have to click on the created methods and format them in >> the browser. >> >> Is there an option to format the method during the compile phase or an >> method that can be passed a collection of methods to format >> automatically? >> >> > I don't know the answer but here is how I might find out.... > Note I am actually only doing this for the first time, so I'll lay it > out in detail in case I can learn from someone else poking holes in it. > > 1. Bring up the context menu in the browser so that it shows the > <Format> menu item > > 2. Use Alt-Shift-Click (on Windows) to bring up halos on the <Format> > menu item, so that you see it is aToggleMenuItemMorph > > 3. Click the green Duplicate action then put the menu item down on the > background. > Check that clicking that copied button will format the code in the browser > > 4. Bring up halos on that duplicated menu item, so that it shows as > aMenuMorph (I don't know why it changed from aToggleMenuItemMorph) > > 5. Click the grey spanner to Debug > Inspect Morph > > 6. Notice that instance variable 'submorphs' is aToggleMenuItemMorph. > Inspect 'submorphs' and then inspect item 1 of the Array to get an > inspector on aToggleMenuItemMorph. > > 7. This inspector shows the following instance variables... > target: anORCmdFormat > selector: #perform:orSendTo: > In the pane at the bottom of the inspector, execute... self selector: > #xxx:yyy: > and observe the instance variable change > > 8. Click the duplicated menuitem still sitting on the background. > A debugger opens with a MNU #xxx:yyy: > Observe that the following message is selected... perform: > selector withArguments: arguments > since in [12.] you will Step Into this point. > > 9. Select the second line from the top of the call stack, which is > #invokeWithEvent: > Observe the instance variable 'selector' shows #xxx:yyy: > > 10. In the bottom pane second from the left, execute... self selector: > #perform:orSendTo: > and observe the instance variable change > > 12. Click <Restart> from the debugger > Step Over until you can Step Into the message you observed in [8.] > After twice stepping-into, you should be in ORCmdFormat>>execute > which looks like it has a few things that will be useful to you. > > I'll leave it there for you to continue tracing. > > cheers -ben >

Jan. 12, 2013

Re: [Pharo-users] Using Utility classes - follow up to How do I save contents of Workspace into a collection?

by Ben Coman

Frank Church wrote: > Igor made this comment in relation to this thread. > > >> indeed. >> I wonder why you (Frank) need such complicated way for doing things. >> If you need a reusable piece of code, which you want to use more than 1 time, >> classes is the best placeholder for them.. >> > > >> Even if you put things at class side, it helps a lot.. so then you can >> just type: >> > > >> MyUtilsClass doThis >> > > >> MyUtilsClass doThat >> > > When I was creating the script to automatically add instance variables > and accessors Stack Overflow 14234003, Mef mentioned that if rather > than run the method against an instance of an object, I could attach > the method to a class, namely ClassDescription. > > >> You could add it to ClassDescription itself, then you wouldn't even have to pass the class as parameter (because self is the class >> then). You could call ClassX addVariablesNamed: aCollection in this case >> > > Is ClassX an example of the MyUtils class Igor mentions? I tried that > option using ClassDescription but got an MNU error.. > > e.g. right now I use the first option in this manner: > > | anyVariable | > anyVariable := ClassType new. > anyVariable addVarsAndTheirAccessors: #('aaaa' 'bbbb' 'cccc') > > How would I attach the method to a utilities class and > ClassDescription for that matter as it would then apply to all > classes? > > I don't quite understand you question, but perhaps executing the following would shed some light... Morph browseHierarchy Morph class browseHierarchy Define the following method... ClassDescription>>mytest self confirm: 'nothing'. Then execute... Morph mytest cheers -ben

Jan. 12, 2013

Re: [Pharo-users] Is there a way to automatically format a method during the compilation phase or immediately after?

by Ben Coman

Frank Church wrote: > I followed the advice from this question on stackoverflow 14234003 and > from there created this simple method to add instance variables and > the basic accessors in one go. > > addVarsAndTheirAccessors: aCollection on: aClass > aCollection do: [ :each | aClass addInstVarNamed: each; > compile: each, ' ^ ', each classified: 'accessing'; > compile: each,': aByteString ',each, ' := aByteString' > classified:'accessing' ] > > Currently I have to click on the created methods and format them in the browser. > > Is there an option to format the method during the compile phase or an > method that can be passed a collection of methods to format > automatically? > > I don't know the answer but here is how I might find out.... Note I am actually only doing this for the first time, so I'll lay it out in detail in case I can learn from someone else poking holes in it. 1. Bring up the context menu in the browser so that it shows the <Format> menu item 2. Use Alt-Shift-Click (on Windows) to bring up halos on the <Format> menu item, so that you see it is aToggleMenuItemMorph 3. Click the green Duplicate action then put the menu item down on the background. Check that clicking that copied button will format the code in the browser 4. Bring up halos on that duplicated menu item, so that it shows as aMenuMorph (I don't know why it changed from aToggleMenuItemMorph) 5. Click the grey spanner to Debug > Inspect Morph 6. Notice that instance variable 'submorphs' is aToggleMenuItemMorph. Inspect 'submorphs' and then inspect item 1 of the Array to get an inspector on aToggleMenuItemMorph. 7. This inspector shows the following instance variables... target: anORCmdFormat selector: #perform:orSendTo: In the pane at the bottom of the inspector, execute... self selector: #xxx:yyy: and observe the instance variable change 8. Click the duplicated menuitem still sitting on the background. A debugger opens with a MNU #xxx:yyy: Observe that the following message is selected... perform: selector withArguments: arguments since in [12.] you will Step Into this point. 9. Select the second line from the top of the call stack, which is #invokeWithEvent: Observe the instance variable 'selector' shows #xxx:yyy: 10. In the bottom pane second from the left, execute... self selector: #perform:orSendTo: and observe the instance variable change 12. Click <Restart> from the debugger Step Over until you can Step Into the message you observed in [8.] After twice stepping-into, you should be in ORCmdFormat>>execute which looks like it has a few things that will be useful to you. I'll leave it there for you to continue tracing. cheers -ben

Jan. 12, 2013