July 2015 - Pharo-dev - lists.pharo.org

Re: [Pharo-dev] [Reflectivity] ... and Roassal

by Tudor Girba

Exactly! This is what happens when all these tools are readily available in the IDE during development. It's often hard to understand the power of these seemingly small things, until you experience their impact on your own problem. Cheers, Doru On Fri, Jul 24, 2015 at 8:57 AM, Marcus Denker <marcus.denker(a)inria.fr> wrote: > Hi, > > Yesterday Alex and me played with using Roassal in combination with > reflectivity. > > Very quickly (within half an hour, even shorter) we had a view per class > that shows coverage for ASTs as trees > like this: > > > > > > this is > > | t | > t := false. > t ifTrue: [ self bar ] ifFalse: [ 'hello' ]. > ^ 5 > > > and you directly see that I have not yet added support for annotating > optimized blocksâ¦ will come soon. > (These views are nice for debugging reflectivityâ¦) > > > The code: > > 1) node-by-node coverage: > > getCoverageOf: aPharoClass whileExecuting: aBlock > " > self new getCoverageOf: CoverageDemoTest whileExecuting: [ > CoverageDemoTest new simpleAst ] > " > > | link | > link := MetaLink new > metaObject: #node; > selector: #tagExecuted. > aPharoClass methods do: [ :cm | cm ast nodesDo: [ :node | node link: link > ] ]. > aBlock value. > ^ aPharoClass > and the code for visalization: > > renderCoverage > | b | > b := RTMondrian new. > > b shape rectangle withTextAbove: #selector. > b nodes: self methods forEach: [ :aMethod | > b shape label > text: [ :astNode | > | aName | > aName := astNode class name copyFrom: 3 to: astNode class name size. > (aName endsWith: 'Node') > ifTrue: [ aName copyFrom: 1 to: aName size - 'Node' size ] > ifFalse: [ aName ] > ] ; > color: Color red; > if: #hasBeenExecuted color: Color green. > b nodes: aMethod ast allChildren. > b edges connectFrom: #parent. > b layout tree. > ]. > > b layout flow. > b build. > ^ b view > > > A second example shows often executed nodes larger (shows another bugâ¦ the > one largest should have the same size as the other large ones): > > > Code: > > > > renderNodeExecutions > | b | > b := RTMondrian new. > > b shape rectangle withTextAbove: #selector. > b nodes: self methods forEach: [ :aMethod | > b shape rectangle > size: [ :astNode | astNode propertyAt: #nbExecutions ifAbsent: [ 0 ] ]. > b nodes: aMethod ast allChildren. > b edges connectFrom: #parent. > b layout tree. > ]. > > b layout flow. > b build. > ^ b view > > > getNodeExecution: aPharoClass whileExecuting: aBlock > " > self new getNodeExecution: CoverageDemoTest whileExecuting: [ > 30 timesRepeat: [ CoverageDemoTest new simpleAst ] ] > " > > | link | > link := MetaLink new > metaObject: [ :node | > | v | > v := node propertyAt: #nbExecutions ifAbsentPut: [ 0 ]. > node propertyAt: #nbExecutions put: v + 1 ]; > selector: #value:; > arguments: #(#node). > aPharoClass methods do: [ :cm | cm ast nodesDo: [ :node | node link: link > ] ]. > aBlock value. > ^ aPharoClass > > > These are just very quick and raw experimentsâ¦ but they got us excited > about the possibilities when this is all > debugged and readyâ¦ > > Marcus > > > > -- www.tudorgirba.com "Every thing has its own flow"

July 24, 2015

[pharo-project/pharo-core]

by GitHub

Branch: refs/tags/50187 Home: https://github.com/pharo-project/pharo-core

July 24, 2015

[pharo-project/pharo-core] 5a3bf1: 50187

by GitHub

Branch: refs/heads/5.0 Home: https://github.com/pharo-project/pharo-core Commit: 5a3bf19d4b2c2cd862dbbb6e9deee2a7c312a722 https://github.com/pharo-project/pharo-core/commit/5a3bf19d4b2c2cd862dbbb6e… Author: Jenkins Build Server <board(a)pharo-project.org> Date: 2015-07-24 (Fri, 24 Jul 2015) Changed paths: R AST-Core.package/RBScanner.class/instance/accessing/flush.st M Kernel.package/CompiledMethod.class/instance/scanning/readsSlot_.st M Kernel.package/Timespan.class/instance/testing/includesAll_.st M Kernel.package/Timespan.class/instance/testing/includesAny_.st R Reflectivity-Tests.package/CoverageDemoTest.class/instance/as yet unclassified/tearDown.st A Reflectivity-Tests.package/CoverageDemoTest.class/instance/running/tearDown.st R Reflectivity-Tests.package/ReflectiveMethodTest.class/instance/as yet unclassified/tearDown.st A Reflectivity-Tests.package/ReflectiveMethodTest.class/instance/running/tearDown.st M Reflectivity.package/Breakpoint.class/instance/links/haltLink.st M Reflectivity.package/Breakpoint.class/instance/links/haltLinkOneShot.st M Reflectivity.package/RFMetaContext.class/instance/level/shiftLevelUp.st R ScriptLoader50.package/ScriptLoader.class/instance/pharo - scripts/script50186.st A ScriptLoader50.package/ScriptLoader.class/instance/pharo - scripts/script50187.st R ScriptLoader50.package/ScriptLoader.class/instance/pharo - updates/update50186.st A ScriptLoader50.package/ScriptLoader.class/instance/pharo - updates/update50187.st M ScriptLoader50.package/ScriptLoader.class/instance/public/commentForCurrentUpdate.st Log Message: ----------- 50187 16019 BreakPoints should install links that compile directly on installation https://pharo.fogbugz.com/f/cases/16019 16017 RBScanner>>#flush not needed https://pharo.fogbugz.com/f/cases/16017 16016 Replace with #allSatisfy:, #anySatisfy: https://pharo.fogbugz.com/f/cases/16016 http://files.pharo.org/image/50/50187.zip

July 24, 2015

[Reflectivity] ... and Roassal

by Marcus Denker

Hi, Yesterday Alex and me played with using Roassal in combination with reflectivity. Very quickly (within half an hour, even shorter) we had a view per class that shows coverage for ASTs as trees like this: this is | t | t := false. t ifTrue: [ self bar ] ifFalse: [ 'hello' ]. ^ 5 and you directly see that I have not yet added support for annotating optimized blocksâ¦ will come soon. (These views are nice for debugging reflectivityâ¦) The code: 1) node-by-node coverage: getCoverageOf: aPharoClass whileExecuting: aBlock " self new getCoverageOf: CoverageDemoTest whileExecuting: [ CoverageDemoTest new simpleAst ] " | link | link := MetaLink new metaObject: #node; selector: #tagExecuted. aPharoClass methods do: [ :cm | cm ast nodesDo: [ :node | node link: link ] ]. aBlock value. ^ aPharoClass and the code for visalization: renderCoverage | b | b := RTMondrian new. b shape rectangle withTextAbove: #selector. b nodes: self methods forEach: [ :aMethod | b shape label text: [ :astNode | | aName | aName := astNode class name copyFrom: 3 to: astNode class name size. (aName endsWith: 'Node') ifTrue: [ aName copyFrom: 1 to: aName size - 'Node' size ] ifFalse: [ aName ] ] ; color: Color red; if: #hasBeenExecuted color: Color green. b nodes: aMethod ast allChildren. b edges connectFrom: #parent. b layout tree. ]. b layout flow. b build. ^ b view A second example shows often executed nodes larger (shows another bugâ¦ the one largest should have the same size as the other large ones): Code: renderNodeExecutions | b | b := RTMondrian new. b shape rectangle withTextAbove: #selector. b nodes: self methods forEach: [ :aMethod | b shape rectangle size: [ :astNode | astNode propertyAt: #nbExecutions ifAbsent: [ 0 ] ]. b nodes: aMethod ast allChildren. b edges connectFrom: #parent. b layout tree. ]. b layout flow. b build. ^ b view getNodeExecution: aPharoClass whileExecuting: aBlock " self new getNodeExecution: CoverageDemoTest whileExecuting: [ 30 timesRepeat: [ CoverageDemoTest new simpleAst ] ] " | link | link := MetaLink new metaObject: [ :node | | v | v := node propertyAt: #nbExecutions ifAbsentPut: [ 0 ]. node propertyAt: #nbExecutions put: v + 1 ]; selector: #value:; arguments: #(#node). aPharoClass methods do: [ :cm | cm ast nodesDo: [ :node | node link: link ] ]. aBlock value. ^ aPharoClass These are just very quick and raw experimentsâ¦ but they got us excited about the possibilities when this is all debugged and readyâ¦ Marcus

July 24, 2015

Re: [Pharo-dev] Slice 16018 add 'Debug tests' menu item to Pharo4 is in the pharo 40 inbox

by Marcus Denker

Hi, Yes, we should put it backâ¦ we just need to find a workaround for the bug. For Pharo5 I am adding it back using breakpoints. (But the machinery is a quite large, so back porting that is not possible). Marcus > On 24 Jul 2015, at 02:19, Paul DeBruicker <pdebruic(a)gmail.com> wrote: > > I added an issue: > > https://pharo.fogbugz.com/f/cases/16018/Add-Debug-test-method-to-Pharo-4-Na… > > The code is in the Pharo40Inbox > > But am not too concerned about it getting into a future release of Pharo 4 because I haven't addressed any of the bugs pointed out in this email thread: > > http://forum.world.st/why-we-removed-debugTest-from-nautilus-td4818657i20.h… > > Because of that I'm not sure it should or should not. But I like the convenience of it & the command works like it did in Pharo 3 and earlier. When using the mouse. I haven't looked into hooking up the keyboard shortcuts. > > Hope this helps someone out. > > Paul

July 24, 2015

Slice 16018 add 'Debug tests' menu item to Pharo4 is in the pharo 40 inbox

by Paul DeBruicker

I added an issue: https://pharo.fogbugz.com/f/cases/16018/Add-Debug-test-method-to-Pharo-4-Na… The code is in the Pharo40Inbox But am not too concerned about it getting into a future release of Pharo 4 because I haven't addressed any of the bugs pointed out in this email thread: http://forum.world.st/why-we-removed-debugTest-from-nautilus-td4818657i20.h… Because of that I'm not sure it should or should not. But I like the convenience of it & the command works like it did in Pharo 3 and earlier. When using the mouse. I haven't looked into hooking up the keyboard shortcuts. Hope this helps someone out. Paul

July 24, 2015

Re: [Pharo-dev] why we removed debugTest from nautilus?

by Paul DeBruicker

Load the Slice #16018 into your Pharo 4 image and it will add the menu item back. Francisco Ortiz PeÃ±aloza wrote > Hi, > > i'm really missing the debug feature in nautilus. +1 to restore it! > > But it's worse than what others describe, lets say you write a tests that > fails, i've to press ALT + "." to've the UI back and the dialog window > that > offers you to proceed, abandon, debug, etc. > > It hangs on this collaboration of > PackageTreeNautilusUI#runTestForAMethod:notifying:priority: > > testSemaphoreForMethod wait. > > I also tried to send #haltOnce or #halt into the test and it doesn't work > on a test. It stops only when i press ALT + "." > > Cheers, > Francisco > > > > > On Sun, Apr 12, 2015 at 7:00 AM, Marcus Denker < > marcus.denker@ > > > wrote: > >> >> > I think break points were done by Clara Allende. As far as I know, they >> work but there's no IDE support for them. I think they are not ready for >> this release but I am not sure about the status. Marcus can confirm it >> will >> not be ready. >> >> For this release the intention was reviewing and integrating the >> Reflectivity machinery, and then afterwards the breakpoints. >> So far Marcus has done a lot on that part (reviewing and enhancing my >> prototype for reflectivity), and we have discussed that the breakpoints >> (in >> the state I've left them at the end of my internship) should work out of >> the box. Nevertheless, I want to do a pass on them and update what's >> necessary once Pharo4 is released. >> >> But they won't be integrated in the release, that's for sure, because >> there's work to do with them... >> >> The problem is that it needs compiler changes that will not be done for >> Pharo4. Maybe we can back port, but the priority for me is not Pharo4 for >> this. >> >> Marcus >> >> -- View this message in context: http://forum.world.st/why-we-removed-debugTest-from-nautilus-tp4818657p4838… Sent from the Pharo Smalltalk Developers mailing list archive at Nabble.com.

July 24, 2015

Re: [Pharo-dev] about quality assistant

by Yuriy Tymchuk

HI, yes, ban may make sense. I think that âfalse positiveâ is a bit incorrect. For example, consider the âprobably missing yourselfâ rule. Even if you are not missing it, the rule is not wrong. It just warns you that you should be carful with this case, because it is easy to screw up something. Now even while this is not a false positive, you want to skip the rule for this method. That was the philosophy behind the name. Uko > On 23 Jul 2015, at 22:29, stepharo <stepharo(a)free.fr> wrote: > > HI yuriy > > it would be good that quality assistant uses the same vocabulary than codecritics. > for example > skip -> ban? > marks as false positif? > > Stef >

July 23, 2015

Re: [Pharo-dev] QualityAssistant in Pharo5

by Yuriy Tymchuk

Would be really nice, but for now Nautilus does not allow to do it. Same goes for changing height when there is only 1 or 2 critics. Either we will improve nautilus after some time, or it will become outrun by other tools. Uko > On 23 Jul 2015, at 21:59, Merwan Ouddane <merwanouddane(a)gmail.com> wrote: > > Great :) > > Can't we make it appear only when he has something to say ? :p > > Le 23/07/2015 18:34, Yuriy Tymchuk a Ã©crit : >> For now there are no summarization features. I.e. When you look at the class you donât see the critics of its methods. As there are no SmallLint rules about packages, I didnât add anything to the package. But this can be an interesting extension in SmallLint at first. >> >> Uko >> >>> On 23 Jul 2015, at 18:15, Peter UhnÃ¡k < <mailto:i.uhnak@gmail.com>i.uhnak(a)gmail.com <mailto:i.uhnak@gmail.com>> wrote: >>> >>> Excellent, thank you! >>> >>> (...and now to fix all offences in my code :)) >>> >>> Also this reminds when I used it some time ago... >>> Can I use it on a package(s) just like Critic Browser? >>> Or maybe since I see it's available in CompiledMethod/TBehavior I could maybe script it... >>> >>> Peter >>> >>> On Thu, Jul 23, 2015 at 5:46 PM, Yuriy Tymchuk < <mailto:yuriy.tymchuk@me.com>yuriy.tymchuk(a)me.com <mailto:yuriy.tymchuk@me.com>> wrote: >>> Dear all, >>> >>> QualityAssistant was integrated in Pharo5. You can read more about its functionality there: https://github.com/Uko/QualityAssistant#quality-assistant- <https://github.com/Uko/QualityAssistant#quality-assistant-> >>> If you will spot any bug please let me know, or report here: https://github.com/Uko/QualityAssistant/issues <https://github.com/Uko/QualityAssistant/issues> >>> >>> Write a good quality code! >>> Uko >>> >> >

July 23, 2015

about quality assistant

by stepharo

HI yuriy it would be good that quality assistant uses the same vocabulary than codecritics. for example skip -> ban? marks as false positif? Stef

July 23, 2015