February 2014 - Pharo-dev - lists.pharo.org

Smalltalkhub is exposing private information
by Norbert Hartl March 3, 2014

March 3, 2014

When you look at someones page you can read commit messages of packages that belong to a private project. You can't see the project nor the package but it is listed in recent activities. I filed a bug https://code.google.com/p/smalltalk-hub/issues/detail?id=32 Norbert

4 3

hashMultiply
by phil＠highoctane.be March 3, 2014

March 3, 2014

I am currently reading through the Hashing in Smalltalk book ( http://www.lulu.com/shop/andres-valloud/hashing-in-smalltalk-theory-and-pra…) and, my head hurting notwithstanding, there are indeed a ton of gems in this system. As he mentions, doing the exercises brings a lot of extra :-) When going to 64-bit, and with the new ObjectMemory scheme, I guess a couple of identity hashing functions will come under scrutiny. e.g. SmallInteger>>hashMultiply | low | low := self bitAnd: 16383. ^(16r260D * low + ((16r260D * (self bitShift: -14) + (16r0065 * low) bitAnd: 16383) * 16384)) bitAnd: 16r0FFFFFFF which will need some more bits. I had a look at how it was done in VisualWorks; hashMultiply "Multiply the receiver by 16r0019660D mod 2^28 without using large integer arithmetic for speed. The constant is a generator of the multiplicative subgroup of Z_2^30, see Knuth's TAOCP vol 2." <primitive: 1747> | low14Bits | low14Bits := self bitAnd: 16r3FFF. ^16384 * (16r260D * (self bitShift: -14) + (16r0065 * low14Bits) bitAnd: 16r3FFF) + (16r260D * low14Bits) bitAnd: 16rFFFFFFF The hashing book version has: multiplication "Computes self times 1664525 mod 2^38 while avoiding overflow into a large integer by making the multiplication into two 14 bits chunks. Do not use any division or modulo operation." | lowBits highBits| lowBits := self bitAnd: 16r3FFF. highBits := self bitShift: -14. ^(lowBits * 16r260D) + (((lowBits * 16r0065) bitAnd: 16r3FFF) bitShift: 14) + (((highBits * 16r260D) bitAnd: 16r3FFF) bitShift: 14) bitAnd: 16rFFFFFFF So, 16384 * is the same as bitShift: 14 and it looks like done once, which may be better. As a side note, how is one debugging such methods? Looks like the debugger (in 2.0) doesn't like SmallIntegers (I can understand why due to the specific nature of the object but still. Ok, Also VW marks it as a primitive, which Pharo does not. Would we gain some speed doing that? hashMultiply is used a lof for identity hashes. Bytecode has quite some work to do: 37 <70> self 38 <20> pushConstant: 16383 39 <BE> send: bitAnd: 40 <68> popIntoTemp: 0 41 <21> pushConstant: 9741 42 <10> pushTemp: 0 43 <B8> send: * 44 <21> pushConstant: 9741 45 <70> self 46 <22> pushConstant: -14 47 <BC> send: bitShift: 48 <B8> send: * 49 <23> pushConstant: 101 50 <10> pushTemp: 0 51 <B8> send: * 52 <B0> send: + 53 <20> pushConstant: 16383 54 <BE> send: bitAnd: 55 <24> pushConstant: 16384 56 <B8> send: * 57 <B0> send: + 58 <25> pushConstant: 268435455 59 <BE> send: bitAnd: 60 <7C> returnTop I ran some experiments timing things. It looks like that replacing 16384 * by bitShift:14 leads to a small gain, bitShift (primitive 17) being faster than * (primitive 9) The bytecode is identical, except send: bitShift instead of send: * multiplication3 | low | low := self bitAnd: 16383. ^(16r260D * low + ((16r260D * (self bitShift: -14) + (16r0065 * low) bitAnd: 16383) bitShift: 14)) bitAnd: 16r0FFFFFFF [500000 timesRepeat: [ 15000 hashMultiply ]] timeToRun 12 [500000 timesRepeat: [ 15000 multiplication ]] timeToRun 41 (worse) [500000 timesRepeat: [ 15000 multiplication3 ]] timeToRun 10 (better) It looks like correct for SmallInteger minVal to: SmallInteger maxVal Now, VW gives: [500000 timesRepeat: [ 15000 hashMultiply ]] timeToRun 1.149 milliseconds Definitely worth investigating the primitive thing, or some NB Asm as this is used about everywhere (Collections etc). Toughts? Phil

3 8

FLMaterializer Materialization error
by Lorenzo Baracchi March 2, 2014

March 2, 2014

Hi I was using FLMaterializer materializationFromFileNamed: filename to read fuel files. Until yesterday everything was fine, but today I updated Pharo and received the error "'Materialization error. Unexpected stream version 19 where it should be 193." I guess that a new version of fuel cannot (simply) read old versions files. Is there a way to know in which version of fuel a file was written and use that version to read it? thanks Lorenzo

4 8

[ANN] Fuel 1.9.3
by Max Leske March 2, 2014

March 2, 2014

Dear fellow Smalltalkers Mariano, Martin and myself are happy to announce Fuel version 1.9.3 which includes the follwing changes to 1.9 (1.9.1 and 1.9.2 were never officially announced): - (feature) the #fuelReplacement selector offers the ability to statically replace an object (e.g. with nil) during analysis. This can lead to significantly smaller graphs and improved speed when serializing large graphs - (feature) serialize the same graph that was analyzed instead of retrazing the graph during serialization. This prevents changes in the graph from happening between analysis and serialization - (change) store source when serializing CompiledMethod objects. Needed because Opal does not allow decompilation - (fix) migrate variables across hierarchy - (feature) Object>>serializeToFileNamed: shortcut for serializing arbitrary objects - (fix) better compression for LargeNegativeIntegers (https://pharo.fogbugz.com/f/cases/12052/Fuel-could-store-LargeNegativeInteg…) Pharo30 already includes most of these changes (integration request: https://pharo.fogbugz.com/f/cases/13000/Integrate-Fuel-1-9-3) This release runs with: - Pharo: 1.1.1, 1.1.2, 1.2, 1.2.2, 1.3, 1.4, 2.0, 3.0 - Squeak: 4.1, 4.2, 4.3, 4.4, 4.5 The documentation is not up to date yet but we will remedy that situatoin within the next couple of days (http://rmod.lille.inria.fr/web/pier/software/Fuel) If you encounter any bugs or have suggestions, please contact us on the Fuel mailing list (http://lists.gforge.inria.fr/cgi-bin/mailman/roster/pharo-fuel) or create a new issue in our bug tracker (https://code.google.com/p/fuel/) We want to thank everyone who contributed to this release (by keyboard or opinion) and also all of those people who actively use Fuel. Cheers, Max

6 7

Help needed! ClassTest>>#testRenaming
by Nicolai Hess March 2, 2014

March 2, 2014

Hope we can sort this out, more and more fixes suffer from the wrong testing failures and nothing new gets integrated. 12996 <https://pharo.fogbugz.com/default.asp?12996>ClassTest>>#testRenaming leaves dirty package Dummy-Tests-Class behind 12995 <https://pharo.fogbugz.com/default.asp?12995> CI ClassTest>>#testRenaming is failing available options: 1. skip test 2. test for nil in RPackageOrganizer>>#systemClassRenamedActionFrom: before rPackage updateDefinedClassNamed: oldName withNewName: newName. 3. review changes for issue 12601 (fix RPackageOrganizer>>#systemCategoryRemovedActionFrom: ann, to unregister packages even if there are MCPackages (don't know about the side effects) best solution would be number 3, but the recent submissions for issue 12601 are quite complex, so I don't know. Nicolai

3 3

ws vs st extensions
by Esteban A. Maringolo March 1, 2014

March 1, 2014

I have a naive question... Why if the BasicCodeLoader handler looks for .st files does the Workpace offer .ws as the default extension? Why not a single extension for Smalltalk files? (.st) and other for chunked content (changesets?) :) Esteban A. Maringolo

4 7

Gettext broken in Pharo 3.0
by Usman Bhatti March 1, 2014

March 1, 2014

Hello, I encountered a problem when trying to export the Gettext file in Pharo 3.0. In a Pharo 3.0 image, load Gettext and then do the following: TextDomainManager registerCategoryPrefix: 'Fuel' domain: 'Fuel'. GetTextExporter exportTemplate. Apparently, there are some deprecations and doing proceed on them shows a DNU: "RBMethodNode does not understand isMessageNode" I can report the bug somewhere to later find some time to correct it but Pharo bug tracker might not be the right place. Any ideas where this bug should be reported? If someone knowledgeable about RB related stuff can have a look at it, it would be great. tx, usman

3 4

#new version #basicNew when materializing (SIXX/Fuel difference)
by Mariano Martinez Peck Feb. 28, 2014

Feb. 28, 2014

Hi guys, I found this a bit strange and arguable. Upon materialization, for normal classes, Fuel uses #basicNew to instantiate objects and then sets its instances variables values getting them from the stream.... Today I found that SIXX sends #new first (*and hence #initialize is sent*) and after, if fails, it tries, #basicNew. This 2 approaches have different results. Which one would you expect? -- Mariano http://marianopeck.wordpress.com

2 1

Pharo 3 shortcuts
by Hernan Wilkinson Feb. 28, 2014

Feb. 28, 2014

Hi, I been using pharo 3 during this week and I'd like to say that it is working great!, it runs faster, looks stable, etc. So, congratulations for the excellent work you are doing. At the same time, I'd like to comment that I feel uncomfortable with the browser's shortcuts and some of them are impossible to use or not consistent. For example, looking for senders on the code panel is alt+n but if you are on the method list it is different, it is alt+b+n which is really difficult to press because three fingers has to be used and the b and n are close one to the other. The one impossible to press is when looking for class references because it is alt+b+N, that is lowercase b and uppercase N at the same time, which is impossible so there is no other solution that to use the menu, but looking for class references is not a first option on the menu but one that is almost the last option on the Analyze menu so you have to navigate a lot with the mouse to get there... For sure there is a reason for these shortcuts... can somebody explain them or point me to where to read them? Thanks -- *HernÃ¡n Wilkinson Agile Software Development, Teaching & Coaching* *Phone: +54 - 011 - *6091 - 3125 * Mobile: +54 - 911 - 4470 - 7207email: hernan.wilkinson(a)10Pines.comsite: http://www.10Pines.com <http://www.10pines.com/>* Address: Alem 693, Floor 5 B, Buenos Aires, Argentina

2 2

git and metacello
by Yuriy Tymchuk Feb. 28, 2014

Feb. 28, 2014

Hi guys! Sorry if Iâm bothering you, but I wasnât able to find any workflow or so on with git. Iâve started to use file-tree, but for now on I donât understand how to make it work with metacello. Also is there some project like that built on jenkins, so I can set up similar job. Cheers Uko

4 15