- Pharo-users - lists.pharo.org

VM dev and compilation
by Bruno Durin Aug. 11, 2017

Aug. 11, 2017

Hi, I have been tinkering with Pharo and Squeak for a few weeks and I am very enthusiastic with these great pieces of software! I would like to ask questions to check my understanding. I am refering from time to time to Squeak as it runs on the same VM as Pharo but my questions are focused on Pharo. As an exercice to learn, I am trying to compile Pharo VMs and get the VM simulator run in a Pharo image (first a 32-bit image+VM and later a 64-bit image+VM). I am doing this both under Ubuntu 14.04 LTS and Mac OS X 10.12.6. What I succeeded in doing: - compiling the squeak and pharo VM using the state of their respective git repo as of about 20 July 2017: on Mac both 32 and 64 bits, on Linux only 64 bits (mainly because I did not want to spend the time to fix the 32-bit libraries which do not coexist well with their 64-bit counterparts), - on Mac, generating the VM development image (SpurVMMaker) for Squeak on top a (32 bits) V5 image, launch it and run the VM simulator, - on Mac, generating the VM development image (generator.image) for Pharo on top of a (32 bits) V5 image, launch it and starting to run a CogVMSimulator (which currently fails, but I am debugging it and learning along the way.) So, my questions: 1) About the Pharo VM compilation (Below "the" Pharo VM means the official VMs that are published on the Pharo website but if VM developers use a different process between 2 releases I would be happy to know the differences.) a) is it correct that the Pharo VM is built using the github pharo-vm repository? b) in the pharo-vm tree, is it correct that "build" and "results" directories are remnant of the build process before Pharo reunited with Squeak on opensmalltalk vm and that the builder classes in the generator.image (PharoVMSpur32Builder and co) do not create the build.sh script in the "build" directory any more. c) is it correct that the Pharo VM is now built with the pharo.cog.spur (for example) make process in the various build sub-directories of the "opensmalltalk-vm" directory? 2) When running a Pharo VM compiled on Mac, the GUI was noticeably slower that the official VM downloaded from pharo.org. (I compared running the official image.) I did not see such slow down when compiling a Squeak VM. My educated guess is that it is related to libcairo and dependencies. Are there compiler optimisations that I should have manually added? Should I generate a XCode project with cmake and compile in XCode (I currently use ./mvm make script)? I have not had this problem under Linux. 3) On Mac, is the difference between having the plugins as Mach-O bundle in Squeak and as dylib libraries in Pharo only a packaging choice or does it have deeper implications? (Does it change for example the way FFI is done?) 4) Do I understand well that currently the simplest way to convert a 32-bit image to a 64-bit one is to use Spur32to64BitBootstrap class in a VM development image? (As far as I understand, the other way is to use SystemTracer.) A remark: when manually compiling a Pharo VM in the pharo-vm repo following Eliot's instructions, the updateSCCSVersions script does not work (because it is one level down with respect to the top of the git tree as compared to the opensmalltalk-vm repository). I eventually understood that the correct build procedure is in the travis scripts but as I do not know well the CI tools, it took a long time before I thought of looking into these scripts. If pharo-vm is used to build the Pharo VM (see question 1.a) I think it would be good to update the build process in README.md in pharo-vm tree (and I can do it). Thanks, Bruno

3 2

UK Smalltalk User Group Meeting - Monday 21st August
by Giovanni Corriga Aug. 10, 2017

Aug. 10, 2017

The next meeting of the UK Smalltalk User Group Meeting will be on Monday, 21st August. We'll meet at a new venue The City Pride ( http://www. cityprideclerkenwell.co.uk/ ) at 7pm. If you'd like to join us, you can just show up at the pub. You can also sign up in advance at the meeting's Meetup page: *https://www.meetup.com/UKSTUG/events/242432390/ <https://www.meetup.com/UKSTUG/events/242432390/> *. See you there!

1 0

about a pillar based output format for Epicea
by Stephane Ducasse Aug. 9, 2017

Aug. 9, 2017

Hi martin I would like to wlak over epicea objects and save the changed methods to be able to write booklets faster. What advices can you give me? Create a special visitor? Stef

4 6

Stylistic question: private helper vs extension method?
by Herby VojÄÃk Aug. 9, 2017

Aug. 9, 2017

Hello! I've got this portion in my delegate: requestPayload ifNotNil: [ uuidKeys do: [ :each | requestPayload at: each ifPresent: [ :s | requestPayload at: each put: (UUID fromString: s) ] ] ]. responsePayload := self towergame clientSync: requestPayload. responsePayload ifNotNil: [ uuidKeys do: [ :each | responsePayload at: each ifPresent: [ :uuid | responsePayload at: each put: uuid asString ] ] ]. Now I would gladly use something like Dictionary >> at:ifPresentTransform: aBlock. But it is not present, so I have two choices: 1. Add it as extension method, but then it may clash if someone else has similar idea. 2. Add private helper TowergameDelegate >> dict:at:ifPresentTransform:, which is longer and needs additional self receiver. Which is preferable? Herby

6 6

API docs as HTML
by Offray Vladimir Luna CÃ¡rdenas Aug. 9, 2017

Aug. 9, 2017

Hi, For the JOSS (Journal of Open Source Software) I need to add to a package documentation the API documentation as an external file (hopefully in HTML). So I would like to export the source code in a way that can be used as API documentation. I remember that some API documentation about Smalltalk/Pharo can be found online, but I don't know how to produce such documents. Any pointers to explore? Cheers, Offray

2 2

Problems verifying the cryptographic signature of a file.
by Offray Vladimir Luna CÃ¡rdenas Aug. 9, 2017

Aug. 9, 2017

Hi, I'm having problems to verify the cryptographic signature of a file. It seems that the test is running, some times, while the file is being created, and some times when the file writing is finished (which is my purpose). How can I run the cryptographic signature verification only after the writing of the file is finished? I imagine that is something related with jobs and completion, but I don't know where to look for, specifically. Any hints and suggestions are welcomed. Cheers, Offray

2 2

gtInspector override default select behavior
by Evan Donahue Aug. 9, 2017

Aug. 9, 2017

Hello, I have a gtInspector extension a tree-structured object. By default, when I click on one of the objects in the gtInspector, the inspector opens up another inspector on that node (using the finder-style display). What I would like to know is, is it possible to override the default click behavior so that, instead of opening the next inspector on that node of the tree, it opens on the result of a block run on that tree. Specifically, I have this: gtInspectorSource: composite <ignoreForCoverage> <gtInspectorPresentationOrder: 10> composite tree title: 'Source'; children: [ :r | r sourceChildren ]; format: [ :r | r sourceName ]; display: self and I was hoping there was some way I could do something like this: ... display: self; onSelectAction: [ :r | r next ] so when I click, I get an inspector not on the node of the tree r, but on r next. Thanks, Evan

2 3

Problem with ThreadPool on os windows - no execution
by Sabine Manaa Aug. 7, 2017

Aug. 7, 2017

Hi, because of [1] I use ThreadPool [2] and while developing on my mac it worked fine. But now, testing it on my production server which is a windows machine, it does not work which means, it is some kind of magic, the code within the block [anObject doSomething] queueWork is not run. When changing it to [anObject halt doSomething] queueWork the code is run! Changing my code from >>queueWork back to >>fork proves that everything is fine now and there is a problem with ThreadPool. When I was playing around to find the problem, I noticed the following: I created a button like this: html mdlButton onClick: (html jQuery ajax script: [ :script | | theSession | theSession := self session. [ theSession inspect ] queueWork. ]); with: 'test' Mac: it works nearly immediately Windows: it does not work ... except when I add a halt in it but: When I then do this after clicking on the test button: 20 timesRepeat: [[ Transcript crShow: Time now printString] queueWork ], the sessions inspector also comes up It seems that the process is hanging until the next work is to be done. Before going deeper in it my question is: does anyone use it on windows? is it made for windows, too? are there any experiences with windows and ThreadPool? did I miss something? how can I fix this? Regards Sabine [1] http://forum.world.st/ThreadPool-with-Seaside-how-td4956812.html#a4957192 [2] http://smalltalkhub.com/#!/~pdebruic/ThreadPool/

2 1

Best Practices for Bundling Documentation
by Evan Donahue Aug. 7, 2017

Aug. 7, 2017

Hello, I'm working on cleaning up a couple of projects for release, and I was wondering what the best practices were for including appropriate documentation with projects. Should I just drop a text blob in a prominent class comment? Is there a tool for structuring documentation? Is documentation typically kept separate from the code? Seems natural to include it so you never have to leave the image, but I'm not sure what most people do. Thanks, Evan

3 7

Lazy-initialization patterns
by Esteban A. Maringolo Aug. 7, 2017

Aug. 7, 2017

Hi there, I have a question about lazy initialization accessor methods that always keep me thinking in terms of performance and style. For years I did something like: MyClass>>#property property isNil ifTrue: [ self initializeProperty ]. "It could be ifNil: as well" ^property MyClass>>#initializeProperty "Returns the receiver, not the property." property := self foo collect: [:e | ... ]. But this made me think that maybe I'm performing extra operations, so I started doing something like: MyClass>>#property ^property ifNil: [ self initializeProperty ] MyClass>>#initializeProperty ^property := self foo collect: [:e | ... ] The initialize method is often created by an "Extract Method" refactoring I do in the accesor, where I do the initial implementation. The messages makes me think that the second implementation is "faster", but an initializationMethod returning the value instead of the receiver is a code smell to me. I early initialize everything I know will be be used right away, but for cases it doesn't make sense. What do you think/prefer? Why? Maybe we could use the reified variables (slots) to implement the lazy initialization themselves :) Best regards! Esteban A. Maringolo

5 6