- Pharo-dev - lists.pharo.org

[Pharo-project] Strange bug in TraitsTest
by Norbert Hartl June 12, 2008

June 12, 2008

If you open a test runner and select only KernelTests-ClassTest and TraitsTests-TraitsTestCase you may find a failure ClassTest>>testRootsOfTheWorld But I don't understand fully why it appears. I had once the chance to get a debugger on the failure (which does usually not work on this failure) and so I know that the rootsOfTheWorld (which is usually 3) is 6 in size containing additionally classes C3 C4 and C6. The ones which are created by TraitsTestCase (see my last mail "Bug for harvest"). They all have a superclass of nil but they aren't included in the SystemNavigation. (conditions for being included in the rootsOfTheWorld) If you press "Run selected" again the same error occurrs while pressing "Run failures" gives you all green (ok, it is this way in my case) And I can't reproduce the same in a 3.10 image. It seems that there is a timing problem involved. Selecting only the two test cases I get the failure every time (or nearly). Selecting all tests it appears only at random. Any hints? Norbert

4 8

[Pharo-project] Fwd: Comment on BrokenTests in pharo
by Pharo Smalltalk June 12, 2008

June 12, 2008

---------- Forwarded message ---------- From: <codesite-noreply(a)google.com> Date: Thu, Jun 12, 2008 at 3:16 AM Subject: Comment on BrokenTests in pharo To: pharo.smalltalk(a)gmail.com Comment by andreas.raab: 3.9 has a different manifest bit pattern for NaN than Pharo. Try this: Float nan hex "<print it>" 3.9: '7FF8000000000000' Pharo: 'FFF8000000000000' It looks like 3.9 was initialized on PPC, Pharo on Intel. Since there is an infinite number of NaN bit patterns the real solution is to use a manifest binary constant in Float initialization, i.e, Float class>>initialize "..." NaN := Float with: 16rFFF80000 with: 0. instead of the computation that is in there today. For more information: http://code.google.com/p/pharo/wiki/BrokenTests

1 0

[Pharo-project] Which tests broken in Phrao are also broken in 3.9 and 3.10
by Oscar Nierstrasz June 12, 2008

June 12, 2008

I checked all the broken tests. It seems that there are some fixes that we should be able to harvest from 3.10. http://code.google.com/p/pharo/wiki/BrokenTests - on

1 0

[Pharo-project] [ANN] New release
by StÃ©phane Ducasse June 12, 2008

June 12, 2008

We are happy to produce a new release. It includes: trait simplification, nebraska removal, MVC removal preparation, tests fixed (float array), semaphore fixes, ... Thanks alain, adrian, norbert, serge and other for their time. Stef Here is the log: update10028 Fixing MC ------------------------------------------------------ update10030 Fixing Delay 0006834: Delay class>>timeoutSemaphore:afterMSecs: doesn t. Description The implementation for Delay class>>timeoutSemaphore:afterMSecs: doesn t even use its first parameter, so it can t do what it says it does. This method should be deprecated or removed entirely. Additional Information + tests Fixing Obsolete references ------------------------------------------------------ withUdpateLog: 1.0 / (FloatArray with: 2.0) answer unexpected result. Bug is in FloatArray>>adaptToNumber:andSend: See : http://bugs.squeak.org/view.php?id=6782 fixing files left by tests ------------------------------------------------------ update10032 Fixing Number parsing Cleaning nebraska ------------------------------------------------------ update10033 Simplifying Traits usage in kernel Simplifying Traits usage in kernel ------------------------------------------------------ update10034 Fixing tests (serge stinckwich) http://bugs.squeak.org/view.php?id=6936 #asHTMLColor is inaccurate for certain values. Attached fix resolves that, also gives 10x speed-up. ------------------------------------------------------ update10035 Merging MVC removal step one of Alain Plantec ------------------------------------------------------ update10036 Merging MVC removal step one of Alain Plantec (missing part)

6 14

[Pharo-project] Trait recent change
by Damien Cassou June 12, 2008

June 12, 2008

Hi, I'm building a new pharo dev and get a message about missing TCommentDescription. Is it normal? -- Damien Cassou Peter von der AhÃ©: Â«I'm beginning to see why Gilad wished us good luckÂ». (http://blogs.sun.com/ahe/entry/override_snafu)

2 2

[Pharo-project] Pharo, squeak and the tickets
by Norbert Hartl June 12, 2008

June 12, 2008

I had a look at the tickets in the google tracker. A lot of them appear to me that they belong to squeak as well. I'm quite unsure what would be the best way to deal with both sides (pharo and squeak) because there are bugs found through working on pharo that belong to squeak as well. On the other hand we harvest things from squeak. What would you prefer? 1. Opening issues where they belong first and let "the others" harvest what they think they need? 2. Playing back the issues to bugs.squeak.org if we find a bug that belongs also to squeak and bring them back by harvesting? How do we want to harvest fixes from other places? 1. Having a wiki page where issues found are written down? 2. Opening tickets on the google tracker for this? This would be "to harvest" tickets. Norbert

4 3

[Pharo-project] KernelTests green
by Oscar Nierstrasz June 12, 2008

June 12, 2008

I have posted some patches to the inbox to make the Kernel Number tests run, and in the Issues page I propose to delete a bunch of nonsensical testHash tests. With these changes the Kernel tests should all be green. - on

4 11

[Pharo-project] Fwd: Comment on BrokenTests in pharo
by Pharo Smalltalk June 12, 2008

June 12, 2008

My apologies if this message is forwarded to the list twice. Victor Rodriguez. ---------- Forwarded message ---------- From: <codesite-noreply(a)google.com> Date: Wed, Jun 11, 2008 at 12:05 PM Subject: Comment on BrokenTests in pharo To: pharo.smalltalk(a)gmail.com Comment by oscar.nierstrasz: FloatTests>>#testNaN5 currently reads as follows: {{{ testNaN5 "self assert: (Float nan asIEEE32BitWord printPaddedWith: $0 to: 32 base: 2) = '01111111110000000000000000000000'." ... }}} I believe this is over-specified. Instead it should just read as follows: {{{ testNaN5 self assert: ((Float nan asIEEE32BitWord printPaddedWith: $0 to: 32 base: 2) copyFrom: 2 to: 9) = '11111111'. ... }}} This would be consistent with the NaN spec (according to wikipedia). However it does not explain why these tests pass in 3.9 and 3.10. Have there been changes in the VM to the way that Floats are represented? For more information: http://code.google.com/p/pharo/wiki/BrokenTests

1 0

[Pharo-project] New pharo-dev based on latest release
by Damien Cassou June 11, 2008

June 11, 2008

I've just published a new pharo-dev based on 10036: https://gforge.inria.fr/frs/download.php/4978/pharo0.1-10036dev08.06.2.zip -- Damien Cassou Peter von der AhÃ©: Â«I'm beginning to see why Gilad wished us good luckÂ». (http://blogs.sun.com/ahe/entry/override_snafu)

3 6

[Pharo-project] HaveALookAt
by Norbert Hartl June 11, 2008

June 11, 2008

I think the link to the packages issue should point to http://www.squeaksource.com/PackagesForSqueak/ Norbert

4 3