June 2025 - Pharo-dev - lists.pharo.org

Pharo News of the Week 26/2025

by Marcus Denker

- [ANN] TinyLogger V2.0.0 https://github.com/jecisc/TinyLogger/releases/tag/v2.0.0 - [ANN] Advanced-Design MOOC on Youtube https://pharo.org/news/2025-06-26-Mooc-Youtube.html - [ANN] Pharo 13.1 Released - https://pharo.org/news/2025-06-26-Pharo13.1.html If you want to add your news to this list, use this form: https://tinyurl.com/4c89buy4 # The week on the Pharo Issue Tracker # Pharo 11 - Implements compatibility with Strict Symbol Comparison feature in Pharo #1682 https://github.com/pharo-spec/Spec/pull/1682 # Pharo 13 - update new tools to v0.10.4 and spec to v2.0.3 #18313 https://github.com/pharo-project/pharo/pull/18313 # Fixes - use UIManager default in OCCodeReparator instead of a direct access to MorphicUIManager #18299 https://github.com/pharo-project/pharo/pull/18299 - Give more feedback in testMethodDefinition #18311 https://github.com/pharo-project/pharo/pull/18311 - Fixed the resize bug and improved code #1180 https://github.com/pharo-spec/NewTools/pull/1180 - Remove fast table search by default #1553 https://github.com/pharo-spec/Spec/pull/1553 # Features - Add shortcut for closing windows #18298 https://github.com/pharo-project/pharo/pull/18298 - Adding trim* API to Text to be polymorphic to String. #18309 https://github.com/pharo-project/pharo/pull/18309 # Tools - small fix on NewTools-Spotter-Processors -> stQuery >> updatefromContext #1146 https://github.com/pharo-spec/NewTools/pull/1146 - Improve Epicea search #18306 https://github.com/pharo-project/pharo/pull/18306 - New methods to define window title for Pull up/down refactorings #18303 https://github.com/pharo-project/pharo/pull/18303 - DrTest second pane displayed as a tree #18297 https://github.com/pharo-project/pharo/pull/18297 - Small fix cavrois window manager #1183 https://github.com/pharo-spec/NewTools/pull/1183 - Fixing new tools baseline #1181 https://github.com/pharo-spec/NewTools/pull/1181 - Simplified whole position back method by using placeholders windows #1179 https://github.com/pharo-spec/NewTools/pull/1179 - Profile windows gets resized with the world #1178 https://github.com/pharo-spec/NewTools/pull/1178 - Added a toggle instead of remove for cell mode #1177 https://github.com/pharo-spec/NewTools/pull/1177 - display the strategies title instead of class name #1176 https://github.com/pharo-spec/NewTools/pull/1176 - Fix browser should not be in "New tools preview" anymore #1171 https://github.com/pharo-spec/NewTools/pull/1171 - Fix StRewriteCritiqueChangesBrowserPresenter opening #1170 https://github.com/pharo-spec/NewTools/pull/1170 - Add stpulse #1164 https://github.com/pharo-spec/NewTools/pull/1164 # Cleanup - rename examples and ClapHelp command #18312 https://github.com/pharo-project/pharo/pull/18312 - Reduced UIManager api #1751 https://github.com/pharo-spec/Spec/pull/1751

June 29, 2025

Re: Confusing Refactoring Pull Up dialog

by stephane ducasse

HI daniel You are right. Now a bit more context of our current work. We are working on this. The point is that the old refactoring code hardcoded in precondition the possible actions in an non user friendly way. Now in Pharo 13 and 14 we have introduced Drivers. A drivers is - using finer decomposition of preconditions to able to propose for example multiple actions such as executing another refactorings to a mere transformation Have a look at Remove Classes on a non empty or classes with subclasses to see what I mean. Now we are adding and reviewing slowly a driver to each refactorings. For the PullUp method the refact > On 29 Jun 2025, at 01:37, Daniel Slomovits <daniels220(a)gmail.com> wrote: > > Yes, I've run into this before. I think the reasoning is that, if you do not copy down the superclass implementation, this is not truly a refactoringâit will change the behavior of a selector that was already understood by the siblings. Indeed this is what the refactoring should do in batch mode. Now in interactive mode then a driver should propose multiple alternatives. > But I agree that in many cases, this change of behavior is the desired outcome and having the option would be good. The simplest change would be to add a "Cancel" button, and make the "No" option do what you wantâperform the refactoring without adding the extra methods. Perhaps adding a warning to the text of the dialog like "Choosing 'No' may modify behavior." (ideally more detailed but I'm having trouble phrasing it concisely). Now to be able to do it we have to rewrite all the conditions to not raise pop up but give enough information to the driver so that it can propose nice choices. S > > On Sat, Jun 28, 2025 at 10:54â¯AM Aik-Siong Koh <askoh(a)askoh.com <mailto:askoh@askoh.com>> wrote: >> <felikbkighlfeoli.png> >> >> There are two questions, but only response accepted. >> I just want a pull up with nothing extra done. >> If I click yes, I get the extra methods. >> If I click no, nothing at all happens. >> >> Please advise, >> Thanks, >> Aik-Siong Koh StÃ©phane Ducasse http://stephane.ducasse.free.fr 06 30 93 66 73 "If you knew today was your last day on earth, what would you do differently? ....ESPECIALLY if, by doing something different, today might not be your last day on earth.â Calvin & Hobbes

June 29, 2025

Re: Does not understand message inconsistency

by Daniel Slomovits

There is a bit of confusion around the word "understand" here. It is sensible to think of an object as "understanding" a message or not, but in fact the selector that tells you whether an object understand a message is #respondsTo:, while the equivalent on the class which tells you whether instances of that class understand is the one that's actually called #canUnderstand:. Other Smalltalk dialects simply say "ReadWriteStream does not understand #selector", no "a[n]" or "Instance of". (I like the use of present tense "does not" instead of past-tense "did not", as well, but that's a much more nitpicky personal preference.) This is in fact already what you are doing with your class-side example, as `PragmaMenuBuilder class` is in fact not the class (that's just `PragmaMenuBuilder`), but the class *of* the class, itself an instance of `Metaclass`. On Fri, Jun 27, 2025 at 4:45â¯PM Aik-Siong Koh <askoh(a)askoh.com> wrote: > "nil did not understand #selector" is great. > > Similarly, "Instance of " can be replace by "a" as in > "aObject did not understand #selector" > "aReadWriteStream did not understand #selector" > > For classes > "PragmaMenuBuilder class did not understand #selector" > > > Daniel Slomovits wrote: > > I think it's because nil *is* pretty special--this is the equivalent of a > NullPointerException or similar in other languages (even though in fact nil > is an object like any other, it *is* usually a different type of *semantic* > error on the part of the programmer than with any other class). > UndefinedObject is also a pretty long class name, so it's nice to leave > that out. But I would be all in favor of making it "nil did not understand > #selector", to be more consistent but still concise. > > On Fri, Jun 27, 2025, 3:00â¯PM Aik-Siong Koh <askoh(a)askoh.com> wrote: > >> Why is UndefinedObject version of "did not understand" so different from >> others? >> It also inverts the object message paradigm. >> Thanks >> Aik-Siong Koh >> >> description >> >> "Returns a textual description of the exception (based on the >> message instance variable). If message is nil, it returns the >> defaultDescription instead." >> >> message ifNil: [ ^ self defaultDescription ]. >> message lookupClass == UndefinedObject ifTrue: [ >> ^ message selector printString , ' was sent to nil' ]. >> >> ^ 'Instance of ' , message lookupClass printString >> , ' did not understand ' , message selector printString >> > >

June 28, 2025

Re: Confusing Refactoring Pull Up dialog

by Daniel Slomovits

Yes, I've run into this before. I think the reasoning is that, if you do not copy down the superclass implementation, this is not truly a refactoringâit will change the behavior of a selector that was already understood by the siblings. But I agree that in many cases, this change of behavior is the desired outcome and having the option would be good. The simplest change would be to add a "Cancel" button, and make the "No" option do what you wantâperform the refactoring without adding the extra methods. Perhaps adding a warning to the text of the dialog like "Choosing 'No' may modify behavior." (ideally more detailed but I'm having trouble phrasing it concisely). On Sat, Jun 28, 2025 at 10:54â¯AM Aik-Siong Koh <askoh(a)askoh.com> wrote: > > > There are two questions, but only response accepted. > I just want a pull up with nothing extra done. > If I click yes, I get the extra methods. > If I click no, nothing at all happens. > > Please advise, > Thanks, > Aik-Siong Koh >

June 28, 2025

Confusing Refactoring Pull Up dialog

by Aik-Siong Koh

There are two questions, but only response accepted. I just want a pull up with nothing extra done. If I click yes, I get the extra methods. If I click no, nothing at all happens. Please advise, Thanks, Aik-Siong Koh

June 28, 2025

Re: Variable references, readers, writers

by Marcus Denker

This is the code that uses assignmentNodes: ((Point slotNamed: #x) assignmentNodes collect: [ :each | each methodNode compiledMethod ] ) asSet But as a downside, it has to create the AST. So maybe better is to use the API of the Variable: Point methods select: [ :each | (Point slotNamed: #x) isWrittenIn: each ] This is implemented to scan on the bytecode level for InstanceVariableSlot and thus is much faster: InstanceVariableSlot>>#isWrittenIn: ^aCompiledCode writesField: index > On 28 Jun 2025, at 00:38, steven.costiou(a)kloum.io wrote: > > Yes, to a variable (so including slots and all) you can ask readNodes and assignmentNodes (accessingNodes gives both). > It gives nodes from the AST, but from there you can obtain the methods and everything. > > Le 27.06.2025 22:27, Aik-Siong Koh a Ã©crit : >> A variable can have references. >> Those references can be split into readers and writers of the variable. >> Can Pharo list readers and writers separately? >> Thanks, >> Aik-Siong Koh

June 28, 2025

Re: Variable references, readers, writers

by steven.costiou＠kloum.io

Yes, to a variable (so including slots and all) you can ask readNodes and assignmentNodes (accessingNodes gives both). It gives nodes from the AST, but from there you can obtain the methods and everything. Le 27.06.2025 22:27, Aik-Siong Koh a Ã©critÂ : > A variable can have references. > Those references can be split into readers and writers of the variable. > Can Pharo list readers and writers separately? > Thanks, > Aik-Siong Koh

June 27, 2025

Re: Does not understand message inconsistency

by Aik-Siong Koh

"nil did not understand #selector" is great. Similarly, "Instance of " can be replace by "a" as in "aObject did not understand #selector" "aReadWriteStream did not understand #selector" For classes "PragmaMenuBuilder class did not understand #selector" Daniel Slomovits wrote: > I think it's because nil *is* pretty special--this is the equivalent > of a NullPointerException or similar in other languages (even though > in fact nil is an object like any other, it *is* usually a different > type of *semantic* error on the part of the programmer than with any > other class). UndefinedObject is also a pretty long class name, so > it's nice to leave that out. But I would be all in favor of making it > "nil did not understand #selector", to be more consistent but still > concise. > > On Fri, Jun 27, 2025, 3:00â¯PM Aik-Siong Koh <askoh(a)askoh.com > <mailto:askoh@askoh.com>> wrote: > > Why is UndefinedObject version of "did not understand" so > different from > others? > It also inverts the object message paradigm. > Thanks > Aik-Siong Koh > > description > > Â Â Â Â "Returns a textual description of the exception (based on the > message instance variable). If message is nil, it returns the > defaultDescription instead." > > Â Â Â Â message ifNil: [ ^ self defaultDescription ]. > Â Â Â Â message lookupClass == UndefinedObject ifTrue: [ > Â Â Â Â Â Â Â ^ message selector printString , ' was sent to nil' ]. > > Â Â Â Â ^ 'Instance of ' , message lookupClass printString > Â Â Â Â Â , ' did not understand ' , message selector printString >

June 27, 2025

Variable references, readers, writers

by Aik-Siong Koh

A variable can have references. Those references can be split into readers and writers of the variable. Can Pharo list readers and writers separately? Thanks, Aik-Siong Koh

June 27, 2025

Re: Does not understand message inconsistency

by Daniel Slomovits

I think it's because nil *is* pretty special--this is the equivalent of a NullPointerException or similar in other languages (even though in fact nil is an object like any other, it *is* usually a different type of *semantic* error on the part of the programmer than with any other class). UndefinedObject is also a pretty long class name, so it's nice to leave that out. But I would be all in favor of making it "nil did not understand #selector", to be more consistent but still concise. On Fri, Jun 27, 2025, 3:00â¯PM Aik-Siong Koh <askoh(a)askoh.com> wrote: > Why is UndefinedObject version of "did not understand" so different from > others? > It also inverts the object message paradigm. > Thanks > Aik-Siong Koh > > description > > "Returns a textual description of the exception (based on the > message instance variable). If message is nil, it returns the > defaultDescription instead." > > message ifNil: [ ^ self defaultDescription ]. > message lookupClass == UndefinedObject ifTrue: [ > ^ message selector printString , ' was sent to nil' ]. > > ^ 'Instance of ' , message lookupClass printString > , ' did not understand ' , message selector printString >

June 27, 2025